Re: [Devel] [Engine-devel] vds_dynamic refactor
On 04/03/2014 07:51 PM, Liran Zelkha wrote:
The problem is with both updates and selects. For selects - to get all the information for the VDS we have multiple joins. Adding another one will hurt performance even more. For updates - we have vds_static thats hardly changed. vds_statistics that changes all the time. vds_dynamic is not changed allot - but is updated all the time because of the status. I think it's best to split it to the two existing tables (BTW - relevant for VM as well)
but we don't update it unless the status has changed, which is a rare occurance?
On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <iheim@redhat.com> wrote:
On 04/03/2014 07:51 PM, Liran Zelkha wrote:
The problem is with both updates and selects. For selects - to get all the information for the VDS we have multiple joins. Adding another one will hurt performance even more. For updates - we have vds_static thats hardly changed. vds_statistics that changes all the time. vds_dynamic is not changed allot - but is updated all the time because of the status. I think it's best to split it to the two existing tables (BTW - relevant for VM as well)
but we don't update it unless the status has changed, which is a rare occurance?
Actually - no. We can definitely see times we are updating vds_dynamic with no reason at all. I tried to create patches for that - but it happens from many different places in the code.
On 04/06/2014 11:32 AM, Liran Zelkha wrote:
On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <iheim@redhat.com <mailto:iheim@redhat.com>> wrote:
On 04/03/2014 07:51 PM, Liran Zelkha wrote:
The problem is with both updates and selects. For selects - to get all the information for the VDS we have multiple joins. Adding another one will hurt performance even more. For updates - we have vds_static thats hardly changed. vds_statistics that changes all the time. vds_dynamic is not changed allot - but is updated all the time because of the status. I think it's best to split it to the two existing tables (BTW - relevant for VM as well)
but we don't update it unless the status has changed, which is a rare occurance?
Actually - no. We can definitely see times we are updating vds_dynamic with no reason at all. I tried to create patches for that - but it happens from many different places in the code.
what would be updated vds_dyanmic for status not originating in update run time info?
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 11:33:12 AM Subject: Re: [Engine-devel] vds_dynamic refactor
On 04/06/2014 11:32 AM, Liran Zelkha wrote:
On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <iheim@redhat.com <mailto:iheim@redhat.com>> wrote:
On 04/03/2014 07:51 PM, Liran Zelkha wrote:
The problem is with both updates and selects. For selects - to get all the information for the VDS we have multiple joins. Adding another one will hurt performance even more. For updates - we have vds_static thats hardly changed. vds_statistics that changes all the time. vds_dynamic is not changed allot - but is updated all the time because of the status. I think it's best to split it to the two existing tables (BTW - relevant for VM as well)
but we don't update it unless the status has changed, which is a rare occurance?
Actually - no. We can definitely see times we are updating vds_dynamic with no reason at all. I tried to create patches for that - but it happens from many different places in the code.
what would be updated vds_dyanmic for status not originating in update run time info?
We have separate DB flows for that (updateStatus and updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). A question: do you know if we update status in updateVdsDynamic? :-) not sure but I found a possible race for pending resources (cpu, mem), LOL :-) Still holds my original thought for having vds_on_boot.
On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 11:33:12 AM Subject: Re: [Engine-devel] vds_dynamic refactor
On 04/06/2014 11:32 AM, Liran Zelkha wrote:
On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <iheim@redhat.com <mailto:iheim@redhat.com>> wrote:
On 04/03/2014 07:51 PM, Liran Zelkha wrote:
The problem is with both updates and selects. For selects - to get all the information for the VDS we have multiple joins. Adding another one will hurt performance even more. For updates - we have vds_static thats hardly changed. vds_statistics that changes all the time. vds_dynamic is not changed allot -
but
is updated all the time because of the status. I think it's best
to
split it to the two existing tables (BTW - relevant for VM as well)
but we don't update it unless the status has changed, which is a rare occurance?
Actually - no. We can definitely see times we are updating vds_dynamic with no reason at all. I tried to create patches for that - but it happens from many different places in the code.
what would be updated vds_dyanmic for status not originating in update run time info?
We have separate DB flows for that (updateStatus and updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). A question: do you know if we update status in updateVdsDynamic? :-) not sure but I found a possible race for pending resources (cpu, mem), LOL :-)
I think we do but not sure. Will check.
Still holds my original thought for having vds_on_boot.
Let's talk f2f on Tuesday?
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Itamar Heim" <iheim@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:26:24 PM Subject: Re: [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 11:33:12 AM Subject: Re: [Engine-devel] vds_dynamic refactor
On 04/06/2014 11:32 AM, Liran Zelkha wrote:
On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <iheim@redhat.com <mailto:iheim@redhat.com>> wrote:
On 04/03/2014 07:51 PM, Liran Zelkha wrote:
The problem is with both updates and selects. For selects - to get all the information for the VDS we have multiple joins. Adding another one will hurt performance even more. For updates - we have vds_static thats hardly changed. vds_statistics that changes all the time. vds_dynamic is not changed allot -
but
is updated all the time because of the status. I think it's best
to
split it to the two existing tables (BTW - relevant for VM as well)
but we don't update it unless the status has changed, which is a rare occurance?
Actually - no. We can definitely see times we are updating vds_dynamic with no reason at all. I tried to create patches for that - but it happens from many different places in the code.
what would be updated vds_dyanmic for status not originating in update run time info?
We have separate DB flows for that (updateStatus and updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). A question: do you know if we update status in updateVdsDynamic? :-) not sure but I found a possible race for pending resources (cpu, mem), LOL :-)
I think we do but not sure. Will check.
Of course it is, that was a rhetorical question :-) (a lot of emoticons and LOLs ;-))
Still holds my original thought for having vds_on_boot.
Let's talk f2f on Tuesday?
I'd prefer to reach conclusions here, I'd like everyone to be involved in a root issue like this one.
Joining in...
From my point of view, in real life a user should have that many VDSs on one Engine (from a DB point of view). Modern DB system handles tables with millions of records and many relations, Do we really have a performance issue here? We could prefer a more easy to maintain implantation in this case over DB performance
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:32:26 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Itamar Heim" <iheim@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:26:24 PM Subject: Re: [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 11:33:12 AM Subject: Re: [Engine-devel] vds_dynamic refactor
On 04/06/2014 11:32 AM, Liran Zelkha wrote:
On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <iheim@redhat.com <mailto:iheim@redhat.com>> wrote:
On 04/03/2014 07:51 PM, Liran Zelkha wrote:
The problem is with both updates and selects. For selects - to get all the information for the VDS we have multiple joins. Adding another one will hurt performance even more. For updates - we have vds_static thats hardly changed. vds_statistics that changes all the time. vds_dynamic is not changed allot -
but
is updated all the time because of the status. I think it's best
to
split it to the two existing tables (BTW - relevant for VM as well)
but we don't update it unless the status has changed, which is a rare occurance?
Actually - no. We can definitely see times we are updating vds_dynamic with no reason at all. I tried to create patches for that - but it happens from many different places in the code.
what would be updated vds_dyanmic for status not originating in update run time info?
We have separate DB flows for that (updateStatus and updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). A question: do you know if we update status in updateVdsDynamic? :-) not sure but I found a possible race for pending resources (cpu, mem), LOL :-)
I think we do but not sure. Will check.
Of course it is, that was a rhetorical question :-) (a lot of emoticons and LOLs ;-))
Still holds my original thought for having vds_on_boot.
Let's talk f2f on Tuesday?
I'd prefer to reach conclusions here, I'd like everyone to be involved in a root issue like this one.
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> wrote:
Joining in... From my point of view, in real life a user should have that many VDSs on one Engine (from a DB point of view). Modern DB system handles tables with millions of records and many relations, Do we really have a performance issue here? We could prefer a more easy to maintain implantation in this case over DB performance
Yes we do. We make many queries on the VDS view, which is a VERY complex view.
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:32:26 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Itamar Heim" <iheim@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:26:24 PM Subject: Re: [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 11:33:12 AM Subject: Re: [Engine-devel] vds_dynamic refactor
On 04/06/2014 11:32 AM, Liran Zelkha wrote:
On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <iheim@redhat.com <mailto:iheim@redhat.com>> wrote:
On 04/03/2014 07:51 PM, Liran Zelkha wrote:
The problem is with both updates and selects. For selects - to get all the information for the VDS we
have
multiple joins. Adding another one will hurt performance even
more.
For updates - we have vds_static thats hardly changed. vds_statistics that changes all the time. vds_dynamic is not changed
allot - but
is updated all the time because of the status. I think it's
best to
split it to the two existing tables (BTW - relevant for VM as
well)
but we don't update it unless the status has changed, which
is a
rare occurance?
Actually - no. We can definitely see times we are updating vds_dynamic with no reason at all. I tried to create patches for that - but
it
happens from many different places in the code.
what would be updated vds_dyanmic for status not originating in update run time info?
We have separate DB flows for that (updateStatus and updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). A question: do you know if we update status in updateVdsDynamic? :-) not sure but I found a possible race for pending resources (cpu, mem), LOL :-)
I think we do but not sure. Will check.
Of course it is, that was a rhetorical question :-) (a lot of emoticons and LOLs ;-))
Still holds my original thought for having vds_on_boot.
Let's talk f2f on Tuesday?
I'd prefer to reach conclusions here, I'd like everyone to be involved in a root issue like this one.
What is the update frequency of this field?
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Kobi Ianko" <kobi@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:40:13 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> wrote:
Joining in... From my point of view, in real life a user should have that many VDSs on one Engine (from a DB point of view). Modern DB system handles tables with millions of records and many relations, Do we really have a performance issue here? We could prefer a more easy to maintain implantation in this case over DB performance
Yes we do. We make many queries on the VDS view, which is a VERY complex view.
Actually I quite agree with Kobi, what is the plan for VMs? why do we start with VDS... what is the biggest deploy do you know of?
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:32:26 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Itamar Heim" <iheim@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:26:24 PM Subject: Re: [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Itamar Heim" <iheim@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 11:33:12 AM Subject: Re: [Engine-devel] vds_dynamic refactor
On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <iheim@redhat.com > <mailto:iheim@redhat.com>> wrote: > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > The problem is with both updates and selects. > For selects - to get all the information for the VDS we have > multiple > joins. Adding another one will hurt performance even more. > For updates - we have vds_static thats hardly changed. > vds_statistics > that changes all the time. vds_dynamic is not changed allot - but > is > updated all the time because of the status. I think it's best to > split > it to the two existing tables (BTW - relevant for VM as well) > > > but we don't update it unless the status has changed, which is a > rare occurance? > > Actually - no. We can definitely see times we are updating > vds_dynamic > with no reason at all. I tried to create patches for that - but it > happens from many different places in the code.
what would be updated vds_dyanmic for status not originating in update run time info?
We have separate DB flows for that (updateStatus and updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). A question: do you know if we update status in updateVdsDynamic? :-) not sure but I found a possible race for pending resources (cpu, mem), LOL :-)
I think we do but not sure. Will check.
Of course it is, that was a rhetorical question :-) (a lot of emoticons and LOLs ;-))
Still holds my original thought for having vds_on_boot.
Let's talk f2f on Tuesday?
I'd prefer to reach conclusions here, I'd like everyone to be involved in a root issue like this one.
What is the update frequency of this field?
which field? status? pending resources? on boot fields? iinm, status is updated mostly by user actions, at least in positive scenarios, and not that often.
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Kobi Ianko" <kobi@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:40:13 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> wrote:
Joining in... From my point of view, in real life a user should have that many VDSs on one Engine (from a DB point of view). Modern DB system handles tables with millions of records and many relations, Do we really have a performance issue here? We could prefer a more easy to maintain implantation in this case over DB performance
Yes we do. We make many queries on the VDS view, which is a VERY complex view.
Actually I quite agree with Kobi, what is the plan for VMs? why do we start with VDS... what is the biggest deploy do you know of?
We start with VDS because in an idle system, with 200 hosts and several thousands VMs, this is what you get as the top queries against the database. Look at how many times getvds is called. [image: Inline image 1] BTW - the second query is an example of abusing the dynamic query mechanism. The 4th query (an update command) is a set of useless update_vds_dynamic commands. For reference, the explain plan of get VDS is something like this: QUERY PLAN ----------------------------------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual time=0.063..0.068 rows=1 loops=1) Join Filter: (vds_static.vds_id = vds_statistics.vds_id) -> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 width=109) (actual time=0.008..0.008 rows=1 loops=1) -> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual time=0.048..0.052 rows=1 loops=1) Join Filter: (vds_groups.vds_group_id = vds_static.vds_group_id) -> Nested Loop Left Join (cost=0.00..9.29 rows=1 width=1389) (actual time=0.013..0.013 rows=1 loops=1) -> Seq Scan on vds_groups (cost=0.00..1.01 rows=1 width=1271) (actual time=0.003..0.003 rows=1 loops=1) -> Index Scan using pk_storage_pool on storage_pool (cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 rows=1 loops=1) Index Cond: (vds_groups.storage_pool_id = id) -> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) (actual time=0.033..0.037 rows=1 loops=1) Hash Cond: (vds_spm_id_map.vds_id = vds_static.vds_id) -> Seq Scan on vds_spm_id_map (cost=0.00..22.30 rows=1230 width=20) (actual time=0.003..0.003 rows=1 loops=1) -> Hash (cost=9.29..9.29 rows=1 width=7606) (actual time=0.019..0.019 rows=1 loops=1) Buckets: 1024 Batches: 1 Memory Usage: 2kB -> Nested Loop (cost=0.00..9.29 rows=1 width=7606) (actual time=0.012..0.013 rows=1 loops=1) -> Seq Scan on vds_dynamic (cost=0.00..1.01 rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1) -> Index Scan using pk_vds_static on vds_static (cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 rows=1 loops=1) Index Cond: (vds_id = vds_dynamic.vds_id) Total runtime: 0.299 ms (19 rows) It's terrible. Adding any additional join will make this worse. Please don't add any more tables...
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org
Sent: Sunday, April 6, 2014 3:32:26 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Itamar Heim" <iheim@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:26:24 PM Subject: Re: [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik <gchaplik@redhat.com
wrote:
----- Original Message ----- > From: "Itamar Heim" <iheim@redhat.com> > To: "Liran Zelkha" <liran.zelkha@gmail.com> > Cc: "Gilad Chaplik" <gchaplik@redhat.com>,
devel@linode01.ovirt.org,
"engine-devel" <engine-devel@ovirt.org> > Sent: Sunday, April 6, 2014 11:33:12 AM > Subject: Re: [Engine-devel] vds_dynamic refactor > > On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim < iheim@redhat.com > > <mailto:iheim@redhat.com>> wrote: > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > > > The problem is with both updates and selects. > > For selects - to get all the information for the VDS we have > > multiple > > joins. Adding another one will hurt performance even more. > > For updates - we have vds_static thats hardly changed. > > vds_statistics > > that changes all the time. vds_dynamic is not changed allot - but > > is > > updated all the time because of the status. I think it's best to > > split > > it to the two existing tables (BTW - relevant for VM as well) > > > > > > but we don't update it unless the status has changed, which is a > > rare occurance? > > > > Actually - no. We can definitely see times we are updating > > vds_dynamic > > with no reason at all. I tried to create patches for that - but it > > happens from many different places in the code. > > what would be updated vds_dyanmic for status not originating in update > run time info?
We have separate DB flows for that (updateStatus and updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). A question: do you know if we update status in updateVdsDynamic? :-) not sure but I found a possible race for pending resources (cpu, mem), LOL :-)
I think we do but not sure. Will check.
Of course it is, that was a rhetorical question :-) (a lot of emoticons and LOLs ;-))
Still holds my original thought for having vds_on_boot.
Let's talk f2f on Tuesday?
I'd prefer to reach conclusions here, I'd like everyone to be involved in a root issue like this one.
What is the update frequency of this field?
which field? status? pending resources? on boot fields? iinm, status is updated mostly by user actions, at least in positive scenarios, and not that often.
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 8:51:02 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Kobi Ianko" <kobi@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:40:13 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> wrote:
Joining in... From my point of view, in real life a user should have that many VDSs on one Engine (from a DB point of view). Modern DB system handles tables with millions of records and many relations, Do we really have a performance issue here? We could prefer a more easy to maintain implantation in this case over DB performance
Yes we do. We make many queries on the VDS view, which is a VERY complex view.
Actually I quite agree with Kobi, what is the plan for VMs? why do we start with VDS... what is the biggest deploy do you know of?
We start with VDS because in an idle system, with 200 hosts and several thousands VMs, this is what you get as the top queries against the database. Look at how many times getvds is called. [image: Inline image 1] BTW - the second query is an example of abusing the dynamic query mechanism. The 4th query (an update command) is a set of useless update_vds_dynamic commands.
For reference, the explain plan of get VDS is something like this:
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual time=0.063..0.068 rows=1 loops=1) Join Filter: (vds_static.vds_id = vds_statistics.vds_id) -> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 width=109) (actual time=0.008..0.008 rows=1 loops=1) -> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual time=0.048..0.052 rows=1 loops=1) Join Filter: (vds_groups.vds_group_id = vds_static.vds_group_id) -> Nested Loop Left Join (cost=0.00..9.29 rows=1 width=1389) (actual time=0.013..0.013 rows=1 loops=1) -> Seq Scan on vds_groups (cost=0.00..1.01 rows=1 width=1271) (actual time=0.003..0.003 rows=1 loops=1) -> Index Scan using pk_storage_pool on storage_pool (cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 rows=1 loops=1) Index Cond: (vds_groups.storage_pool_id = id) -> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) (actual time=0.033..0.037 rows=1 loops=1) Hash Cond: (vds_spm_id_map.vds_id = vds_static.vds_id) -> Seq Scan on vds_spm_id_map (cost=0.00..22.30 rows=1230 width=20) (actual time=0.003..0.003 rows=1 loops=1) -> Hash (cost=9.29..9.29 rows=1 width=7606) (actual time=0.019..0.019 rows=1 loops=1) Buckets: 1024 Batches: 1 Memory Usage: 2kB -> Nested Loop (cost=0.00..9.29 rows=1 width=7606) (actual time=0.012..0.013 rows=1 loops=1) -> Seq Scan on vds_dynamic (cost=0.00..1.01 rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1) -> Index Scan using pk_vds_static on vds_static (cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 rows=1 loops=1) Index Cond: (vds_id = vds_dynamic.vds_id) Total runtime: 0.299 ms (19 rows)
It's terrible. Adding any additional join will make this worse. Please don't add any more tables...
Thank you for the detailed explanation, my comments: * a very long time isn't an argument for not adding another table (should be neglectable); currently we have an unrelated problem, we need to solve it. * > We start with VDS because in an idle system, with 200 hosts and several
thousands VMs, this is what you get as the top queries against the database.
so, if fetching VMs takes 10 minutes? and its get called a single time? * you didn't reply on my of my suggestion of constructing the VDS records in the DB without using joins.
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org
Sent: Sunday, April 6, 2014 3:32:26 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Itamar Heim" <iheim@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:26:24 PM Subject: Re: [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik <gchaplik@redhat.com
wrote:
> ----- Original Message ----- > > From: "Itamar Heim" <iheim@redhat.com> > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>,
devel@linode01.ovirt.org,
> "engine-devel" <engine-devel@ovirt.org> > > Sent: Sunday, April 6, 2014 11:33:12 AM > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > > > > > > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim < iheim@redhat.com > > > <mailto:iheim@redhat.com>> wrote: > > > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > > > > > The problem is with both updates and selects. > > > For selects - to get all the information for the VDS we have > > > multiple > > > joins. Adding another one will hurt performance even more. > > > For updates - we have vds_static thats hardly changed. > > > vds_statistics > > > that changes all the time. vds_dynamic is not changed allot - > but > > > is > > > updated all the time because of the status. I think it's best > to > > > split > > > it to the two existing tables (BTW - relevant for VM as well) > > > > > > > > > but we don't update it unless the status has changed, which is a > > > rare occurance? > > > > > > Actually - no. We can definitely see times we are updating > > > vds_dynamic > > > with no reason at all. I tried to create patches for that - but it > > > happens from many different places in the code. > > > > what would be updated vds_dyanmic for status not originating in update > > run time info? > > We have separate DB flows for that (updateStatus and > updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). > A question: do you know if we update status in updateVdsDynamic? :-) not > sure but I found a possible race for pending resources (cpu, mem), LOL > :-) > > I think we do but not sure. Will check.
Of course it is, that was a rhetorical question :-) (a lot of emoticons and LOLs ;-))
> Still holds my original thought for having vds_on_boot. > > > Let's talk f2f on Tuesday?
I'd prefer to reach conclusions here, I'd like everyone to be involved in a root issue like this one.
What is the update frequency of this field?
which field? status? pending resources? on boot fields? iinm, status is updated mostly by user actions, at least in positive scenarios, and not that often.
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
On Sun, Apr 6, 2014 at 11:03 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 8:51:02 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Kobi Ianko" <kobi@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:40:13 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> wrote:
Joining in... From my point of view, in real life a user should have that many VDSs on one Engine (from a DB point of view). Modern DB system handles tables with millions of records and many relations, Do we really have a performance issue here? We could prefer a more easy to maintain implantation in this case over DB performance
Yes we do. We make many queries on the VDS view, which is a VERY complex view.
Actually I quite agree with Kobi, what is the plan for VMs? why do we start with VDS... what is the biggest deploy do you know of?
We start with VDS because in an idle system, with 200 hosts and several thousands VMs, this is what you get as the top queries against the database. Look at how many times getvds is called. [image: Inline image 1] BTW - the second query is an example of abusing the dynamic query mechanism. The 4th query (an update command) is a set of useless update_vds_dynamic commands.
For reference, the explain plan of get VDS is something like this:
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------------
Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual time=0.063..0.068 rows=1 loops=1) Join Filter: (vds_static.vds_id = vds_statistics.vds_id) -> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 width=109) (actual time=0.008..0.008 rows=1 loops=1) -> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual time=0.048..0.052 rows=1 loops=1) Join Filter: (vds_groups.vds_group_id = vds_static.vds_group_id) -> Nested Loop Left Join (cost=0.00..9.29 rows=1 width=1389) (actual time=0.013..0.013 rows=1 loops=1) -> Seq Scan on vds_groups (cost=0.00..1.01 rows=1 width=1271) (actual time=0.003..0.003 rows=1 loops=1) -> Index Scan using pk_storage_pool on storage_pool (cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 rows=1 loops=1) Index Cond: (vds_groups.storage_pool_id = id) -> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) (actual time=0.033..0.037 rows=1 loops=1) Hash Cond: (vds_spm_id_map.vds_id = vds_static.vds_id) -> Seq Scan on vds_spm_id_map (cost=0.00..22.30 rows=1230 width=20) (actual time=0.003..0.003 rows=1 loops=1) -> Hash (cost=9.29..9.29 rows=1 width=7606) (actual time=0.019..0.019 rows=1 loops=1) Buckets: 1024 Batches: 1 Memory Usage: 2kB -> Nested Loop (cost=0.00..9.29 rows=1 width=7606) (actual time=0.012..0.013 rows=1 loops=1) -> Seq Scan on vds_dynamic (cost=0.00..1.01 rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1) -> Index Scan using pk_vds_static on vds_static (cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 rows=1 loops=1) Index Cond: (vds_id = vds_dynamic.vds_id) Total runtime: 0.299 ms (19 rows)
It's terrible. Adding any additional join will make this worse. Please don't add any more tables...
Thank you for the detailed explanation, my comments:
* a very long time isn't an argument for not adding another table (should be neglectable); currently we have an unrelated problem, we need to solve it.
Of course it is. A very long time for a query that you execute many times is THE factor. Who said the join has no performance effect? Have you tested it? Under load? Under many writes/updates?
* > We start with VDS because in an idle system, with 200 hosts and several
thousands VMs, this is what you get as the top queries against the database.
so, if fetching VMs takes 10 minutes? and its get called a single time?
Where do you see 10 minutes? If you are looking at the red bar it's the inherent time - total query time * number of queries.
* you didn't reply on my of my suggestion of constructing the VDS records in the DB without using joins.
If you mean materialized views - we don't have it in Postgres just yet... And even if we do, since we do many updates to vds_statistics and vds_dynamic - I'm not sure it will have positive impact on our performance. If you mean joins in the database - everything that is based on VDS is done in the database. Part of the problem, since we can cache some information and only query the dynamic/statistics part of VDS, but that's another matter.
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: devel@linode01.ovirt.org, "engine-devel" <
Sent: Sunday, April 6, 2014 3:32:26 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message ----- > From: "Liran Zelkha" <liran.zelkha@gmail.com> > To: "Gilad Chaplik" <gchaplik@redhat.com> > Cc: "Itamar Heim" <iheim@redhat.com>, devel@linode01.ovirt.org
,
> "engine-devel" <engine-devel@ovirt.org> > Sent: Sunday, April 6, 2014 3:26:24 PM > Subject: Re: [Engine-devel] vds_dynamic refactor > > On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik < gchaplik@redhat.com
wrote:
> > > ----- Original Message ----- > > > From: "Itamar Heim" <iheim@redhat.com> > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, > > "engine-devel" <engine-devel@ovirt.org> > > > Sent: Sunday, April 6, 2014 11:33:12 AM > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > > > > > > > > > > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim < iheim@redhat.com > > > > <mailto:iheim@redhat.com>> wrote: > > > > > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > > > > > > > The problem is with both updates and selects. > > > > For selects - to get all the information for the VDS we have > > > > multiple > > > > joins. Adding another one will hurt performance even more. > > > > For updates - we have vds_static thats hardly changed. > > > > vds_statistics > > > > that changes all the time. vds_dynamic is not changed allot - > > but > > > > is > > > > updated all the time because of the status. I
it's
best
> > to > > > > split > > > > it to the two existing tables (BTW - relevant for VM as well) > > > > > > > > > > > > but we don't update it unless the status has changed, which is a > > > > rare occurance? > > > > > > > > Actually - no. We can definitely see times we are updating > > > > vds_dynamic > > > > with no reason at all. I tried to create patches for
engine-devel@ovirt.org think that -
but
it
> > > > happens from many different places in the code. > > > > > > what would be updated vds_dyanmic for status not originating in update > > > run time info? > > > > We have separate DB flows for that (updateStatus and > > updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). > > A question: do you know if we update status in updateVdsDynamic? :-) not > > sure but I found a possible race for pending resources (cpu, mem), LOL > > :-) > > > > I think we do but not sure. Will check.
Of course it is, that was a rhetorical question :-) (a lot of emoticons and LOLs ;-))
> > > > Still holds my original thought for having vds_on_boot. > > > > > > > Let's talk f2f on Tuesday?
I'd prefer to reach conclusions here, I'd like everyone to be involved in a root issue like this one.
What is the update frequency of this field?
which field? status? pending resources? on boot fields? iinm, status is updated mostly by user actions, at least in positive scenarios, and not that often.
> _______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
------=_Part_3278465_796563103.1397046843672 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Why not move only status with changes a lot to statistics and leave everything as is? Yaniv ----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Monday, April 7, 2014 8:51:00 AM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 11:03 PM, Gilad Chaplik < gchaplik@redhat.com > wrote:
----- Original Message -----
From: "Liran Zelkha" < liran.zelkha@gmail.com >
To: "Gilad Chaplik" < gchaplik@redhat.com >
Cc: "Kobi Ianko" < kobi@redhat.com >, devel@linode01.ovirt.org , "engine-devel" < engine-devel@ovirt.org >
Sent: Sunday, April 6, 2014 8:51:02 PM
Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik < gchaplik@redhat.com > wrote:
----- Original Message -----
From: "Liran Zelkha" < liran.zelkha@gmail.com >
To: "Kobi Ianko" < kobi@redhat.com >
Cc: "Gilad Chaplik" < gchaplik@redhat.com >, devel@linode01.ovirt.org ,
"engine-devel" < engine-devel@ovirt.org >
Sent: Sunday, April 6, 2014 3:40:13 PM
Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko < kobi@redhat.com > wrote:
Joining in...
From my point of view, in real life a user should have that many VDSs
on
one Engine (from a DB point of view).
Modern DB system handles tables with millions of records and many
relations, Do we really have a performance issue here?
We could prefer a more easy to maintain implantation in this case over
DB
performance
Yes we do. We make many queries on the VDS view, which is a VERY
complex
view.
Actually I quite agree with Kobi, what is the plan for VMs? why do we
start with VDS...
what is the biggest deploy do you know of?
We start with VDS because in an idle system, with 200 hosts and several
thousands VMs, this is what you get as the top queries against the
database. Look at how many times getvds is called.
[image: Inline image 1]
BTW - the second query is an example of abusing the dynamic query
mechanism. The 4th query (an update command) is a set of useless
update_vds_dynamic commands.
For reference, the explain plan of get VDS is something like this:
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------------
Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual
time=0.063..0.068 rows=1 loops=1)
Join Filter: (vds_static.vds_id = vds_statistics.vds_id)
-> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 width=109)
(actual time=0.008..0.008 rows=1 loops=1)
-> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual
time=0.048..0.052 rows=1 loops=1)
Join Filter: (vds_groups.vds_group_id = vds_static.vds_group_id)
-> Nested Loop Left Join (cost=0.00..9.29 rows=1 width=1389)
(actual time=0.013..0.013 rows=1 loops=1)
-> Seq Scan on vds_groups (cost=0.00..1.01 rows=1
width=1271) (actual time=0.003..0.003 rows=1 loops=1)
-> Index Scan using pk_storage_pool on storage_pool
(cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 rows=1
loops=1)
Index Cond: (vds_groups.storage_pool_id = id)
-> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) (actual
time=0.033..0.037 rows=1 loops=1)
Hash Cond: (vds_spm_id_map.vds_id = vds_static.vds_id)
-> Seq Scan on vds_spm_id_map (cost=0.00..22.30 rows=1230
width=20) (actual time=0.003..0.003 rows=1 loops=1)
-> Hash (cost=9.29..9.29 rows=1 width=7606) (actual
time=0.019..0.019 rows=1 loops=1)
Buckets: 1024 Batches: 1 Memory Usage: 2kB
-> Nested Loop (cost=0.00..9.29 rows=1 width=7606)
(actual time=0.012..0.013 rows=1 loops=1)
-> Seq Scan on vds_dynamic (cost=0.00..1.01
rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1)
-> Index Scan using pk_vds_static on vds_static
(cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 rows=1
loops=1)
Index Cond: (vds_id = vds_dynamic.vds_id)
Total runtime: 0.299 ms
(19 rows)
It's terrible. Adding any additional join will make this worse. Please
don't add any more tables...
Thank you for the detailed explanation, my comments:
* a very long time isn't an argument for not adding another table (should be neglectable);
currently we have an unrelated problem, we need to solve it.
Of course it is. A very long time for a query that you execute many times is THE factor. Who said the join has no performance effect? Have you tested it? Under load? Under many writes/updates?
* > We start with VDS because in an idle system, with 200 hosts and several
thousands VMs, this is what you get as the top queries against the
database.
so, if fetching VMs takes 10 minutes? and its get called a single time?
Where do you see 10 minutes? If you are looking at the red bar it's the inherent time - total query time * number of queries.
* you didn't reply on my of my suggestion of constructing the VDS records in the DB without using joins.
If you mean materialized views - we don't have it in Postgres just yet... And even if we do, since we do many updates to vds_statistics and vds_dynamic - I'm not sure it will have positive impact on our performance. If you mean joins in the database - everything that is based on VDS is done in the database. Part of the problem, since we can cache some information and only query the dynamic/statistics part of VDS, but that's another matter.
----- Original Message -----
> From: "Gilad Chaplik" < gchaplik@redhat.com >
> To: "Liran Zelkha" < liran.zelkha@gmail.com >
> Cc: devel@linode01.ovirt.org , "engine-devel" < > engine-devel@ovirt.org
> Sent: Sunday, April 6, 2014 3:32:26 PM
> Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
>
> ----- Original Message -----
> > From: "Liran Zelkha" < liran.zelkha@gmail.com >
> > To: "Gilad Chaplik" < gchaplik@redhat.com >
> > Cc: "Itamar Heim" < iheim@redhat.com >, > > devel@linode01.ovirt.org > > ,
> > "engine-devel" < engine-devel@ovirt.org >
> > Sent: Sunday, April 6, 2014 3:26:24 PM
> > Subject: Re: [Engine-devel] vds_dynamic refactor
> >
> > On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik < > > gchaplik@redhat.com
wrote:
> >
> > > ----- Original Message -----
> > > > From: "Itamar Heim" < iheim@redhat.com >
> > > > To: "Liran Zelkha" < liran.zelkha@gmail.com >
> > > > Cc: "Gilad Chaplik" < gchaplik@redhat.com >,
devel@linode01.ovirt.org ,
> > > "engine-devel" < engine-devel@ovirt.org >
> > > > Sent: Sunday, April 6, 2014 11:33:12 AM
> > > > Subject: Re: [Engine-devel] vds_dynamic refactor
> > > >
> > > > On 04/06/2014 11:32 AM, Liran Zelkha wrote:
> > > > >
> > > > >
> > > > >
> > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim <
iheim@redhat.com
> > > > > <mailto: iheim@redhat.com >> wrote:
> > > > >
> > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote:
> > > > >
> > > > > The problem is with both updates and selects.
> > > > > For selects - to get all the information for the VDS
we
have
> > > > > multiple
> > > > > joins. Adding another one will hurt performance even
more.
> > > > > For updates - we have vds_static thats hardly
changed.
> > > > > vds_statistics
> > > > > that changes all the time. vds_dynamic is not changed
allot -
> > > but
> > > > > is
> > > > > updated all the time because of the status. I think
it's
best
> > > to
> > > > > split
> > > > > it to the two existing tables (BTW - relevant for VM
as
well)
> > > > >
> > > > >
> > > > > but we don't update it unless the status has changed,
which
is a
> > > > > rare occurance?
> > > > >
> > > > > Actually - no. We can definitely see times we are > > > > > updating
> > > > > vds_dynamic
> > > > > with no reason at all. I tried to create patches for that > > > > > -
but
it
> > > > > happens from many different places in the code.
> > > >
> > > > what would be updated vds_dyanmic for status not > > > > originating > > > > in
update
> > > > run time info?
> > >
> > > We have separate DB flows for that (updateStatus and
> > > updatePartialVdsDynamicCalc and more in
VdsDynamicDAODbFacadeImpl).
> > > A question: do you know if we update status in > > > updateVdsDynamic?
:-)
not
> > > sure but I found a possible race for pending resources (cpu,
mem),
LOL
> > > :-)
> > >
> > > I think we do but not sure. Will check.
>
> Of course it is, that was a rhetorical question :-) (a lot of
emoticons
and
> LOLs ;-))
>
> >
> >
> > > Still holds my original thought for having vds_on_boot.
> > >
> > >
> > >
> > Let's talk f2f on Tuesday?
>
> I'd prefer to reach conclusions here, I'd like everyone to be
involved
in a
> root issue like this one.
What is the update frequency of this field?
which field?
status? pending resources? on boot fields?
iinm, status is updated mostly by user actions, at least in positive
scenarios, and not that often.
>
> >
> _______________________________________________
> Devel mailing list
> Devel@ovirt.org
>
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel ------=_Part_3278465_796563103.1397046843672 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable
<br> > > ><br> > > > > > Sent: Sunday, April 6, 2014 3:32:26 PM<br> > > > > > Subject: Re: [Devel] [Engine-devel] vds_dynamic re= factor<br> > > > > ><br> > > > > > ----- Original Message -----<br> > > > > > > From: "Liran Zelkha" <<a href=3D"mailto:li= ran.zelkha@gmail.com" target=3D"_blank">liran.zelkha@gmail.com</a>><br> > > > > > > To: "Gilad Chaplik" <<a href=3D"mailto:gch= aplik@redhat.com" target=3D"_blank">gchaplik@redhat.com</a>><br> > > > > > > Cc: "Itamar Heim" <<a href=3D"mailto:iheim= @redhat.com" target=3D"_blank">iheim@redhat.com</a>>, <a href=3D"mailto:= devel@linode01.ovirt.org" target=3D"_blank">devel@linode01.ovirt.org</a>,<b= r> > > > > > > "engine-devel" <<a href=3D"mailto:engine-d= evel@ovirt.org" target=3D"_blank">engine-devel@ovirt.org</a>><br> > > > > > > Sent: Sunday, April 6, 2014 3:26:24 PM<br> > > > > > > Subject: Re: [Engine-devel] vds_dynamic refac= tor<br> > > > > > ><br> > > > > > > On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik= <<a href=3D"mailto:gchaplik@redhat.com" target=3D"_blank">gchaplik@redh= at.com</a><br> > > ><br> > > > > wrote:<br> > > > > > ><br> > > > > > > > ----- Original Message -----<br> > > > > > > > > From: "Itamar Heim" <<a href=3D"= mailto:iheim@redhat.com" target=3D"_blank">iheim@redhat.com</a>><br> > > > > > > > > To: "Liran Zelkha" <<a href=3D"m= ailto:liran.zelkha@gmail.com" target=3D"_blank">liran.zelkha@gmail.com</a>&= gt;<br> > > > > > > > > Cc: "Gilad Chaplik" <<a href=3D"= mailto:gchaplik@redhat.com" target=3D"_blank">gchaplik@redhat.com</a>>,<= br> > > > > <a href=3D"mailto:devel@linode01.ovirt.org" target=3D"_= blank">devel@linode01.ovirt.org</a>,<br> > > > > > > > "engine-devel" <<a href=3D"mailto:eng= ine-devel@ovirt.org" target=3D"_blank">engine-devel@ovirt.org</a>><br> > > > > > > > > Sent: Sunday, April 6, 2014 11:33:1= 2 AM<br> > > > > > > > > Subject: Re: [Engine-devel] vds_dyn= amic refactor<br> > > > > > > > ><br> > > > > > > > > On 04/06/2014 11:32 AM, Liran Zelkh= a wrote:<br> > > > > > > > > ><br> > > > > > > > > ><br> > > > > > > > > ><br> > > > > > > > > > On Sun, Apr 6, 2014 at 11:22 A= M, Itamar Heim <<br> > > <a href=3D"mailto:iheim@redhat.com" target=3D"_blank">iheim@redha= t.com</a><br> > > > > > > > > > <mailto:<a href=3D"mailto:i= heim@redhat.com" target=3D"_blank">iheim@redhat.com</a>>> wrote:<br> > > > > > > > > ><br> > > > > > > > > > On 04/03/2014 07= :51 PM, Liran Zelkha wrote:<br> > > > > > > > > ><br> > > > > > > > > > Th= e problem is with both updates and selects.<br> > > > > > > > > > Fo= r selects - to get all the information for the VDS<br> > > we<br> > > > > have<br> > > > > > > > > > mu= ltiple<br> > > > > > > > > > jo= ins. Adding another one will hurt performance even<br> > > > > more.<br> > > > > > > > > > Fo= r updates - we have vds_static thats hardly<br> > > changed.<br> > > > > > > > > > vd= s_statistics<br> > > > > > > > > > th= at changes all the time. vds_dynamic is not changed<br> > > > > allot -<br> > > > > > > > but<br> > > > > > > > > > is= <br> > > > > > > > > > up= dated all the time because of the status. I think<br> > > it's<br> > > > > best<br> > > > > > > > to<br> > > > > > > > > > sp=
<html><body><div style=3D"font-family: Arial; font-size: 10pt; color: #0000= 00"><div>Why not move only status with changes a lot to statistics and leav= e everything as is?<br></div><div><br></div><div><br></div><div><br></div><= div><br></div><div>Yaniv</div><div><br></div><hr id=3D"zwchr"><blockquote s= tyle=3D"border-left:2px solid #1010FF;margin-left:5px;padding-left:5px;colo= r:#000;font-weight:normal;font-style:normal;text-decoration:none;font-famil= y:Helvetica,Arial,sans-serif;font-size:12pt;"><b>From: </b>"Liran Zelkha" &= lt;liran.zelkha@gmail.com><br><b>To: </b>"Gilad Chaplik" <gchaplik@re= dhat.com><br><b>Cc: </b>"Kobi Ianko" <kobi@redhat.com>, devel@lino= de01.ovirt.org, "engine-devel" <engine-devel@ovirt.org><br><b>Sent: <= /b>Monday, April 7, 2014 8:51:00 AM<br><b>Subject: </b>Re: [Devel] [Engine-= devel] vds_dynamic refactor<br><div><br></div><div dir=3D"ltr"><br><div cla= ss=3D"gmail_extra"><br><div><br></div><div class=3D"gmail_quote">On Sun, Ap= r 6, 2014 at 11:03 PM, Gilad Chaplik <span dir=3D"ltr"><<a href=3D"mailt= o:gchaplik@redhat.com" target=3D"_blank">gchaplik@redhat.com</a>></span>= wrote:<br> <blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p= x #ccc solid;padding-left:1ex"><div class=3D"">----- Original Message -----= <br> > From: "Liran Zelkha" <<a href=3D"mailto:liran.zelkha@gmail.com" tar= get=3D"_blank">liran.zelkha@gmail.com</a>><br> > To: "Gilad Chaplik" <<a href=3D"mailto:gchaplik@redhat.com" target= =3D"_blank">gchaplik@redhat.com</a>><br> </div><div><div class=3D"h5">> Cc: "Kobi Ianko" <<a href=3D"mailto:ko= bi@redhat.com" target=3D"_blank">kobi@redhat.com</a>>, <a href=3D"mailto= :devel@linode01.ovirt.org" target=3D"_blank">devel@linode01.ovirt.org</a>, = "engine-devel" <<a href=3D"mailto:engine-devel@ovirt.org" target=3D"_bla= nk">engine-devel@ovirt.org</a>><br> > Sent: Sunday, April 6, 2014 8:51:02 PM<br> > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor<br> ><br> > On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik <<a href=3D"mailto:gc= haplik@redhat.com" target=3D"_blank">gchaplik@redhat.com</a>> wrote:<br> ><br> > > ----- Original Message -----<br> > > > From: "Liran Zelkha" <<a href=3D"mailto:liran.zelkha@gmai= l.com" target=3D"_blank">liran.zelkha@gmail.com</a>><br> > > > To: "Kobi Ianko" <<a href=3D"mailto:kobi@redhat.com" targ= et=3D"_blank">kobi@redhat.com</a>><br> > > > Cc: "Gilad Chaplik" <<a href=3D"mailto:gchaplik@redhat.co= m" target=3D"_blank">gchaplik@redhat.com</a>>, <a href=3D"mailto:devel@l= inode01.ovirt.org" target=3D"_blank">devel@linode01.ovirt.org</a>,<br> > > "engine-devel" <<a href=3D"mailto:engine-devel@ovirt.org" targ= et=3D"_blank">engine-devel@ovirt.org</a>><br> > > > Sent: Sunday, April 6, 2014 3:40:13 PM<br> > > > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor<br> > > ><br> > > > On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <<a href=3D"ma= ilto:kobi@redhat.com" target=3D"_blank">kobi@redhat.com</a>> wrote:<br> > > ><br> > > > > Joining in...<br> > > > > From my point of view, in real life a user should have = that many VDSs<br> > > on<br> > > > > one Engine (from a DB point of view).<br> > > > > Modern DB system handles tables with millions of record= s and many<br> > > > > relations, Do we really have a performance issue here?<= br> > > > > We could prefer a more easy to maintain implantation in= this case over<br> > > DB<br> > > > > performance<br> > > > ><br> > > > > Yes we do. We make many queries on the VDS view, which = is a VERY<br> > > complex<br> > > > view.<br> > > ><br> > ><br> > > Actually I quite agree with Kobi, what is the plan for VMs? why d= o we<br> > > start with VDS...<br> > > what is the biggest deploy do you know of?<br> > ><br> > We start with VDS because in an idle system, with 200 hosts and severa= l<br> > thousands VMs, this is what you get as the top queries against the<br> > database. Look at how many times getvds is called.<br> > [image: Inline image 1]<br> > BTW - the second query is an example of abusing the dynamic query<br> > mechanism. The 4th query (an update command) is a set of useless<br> > update_vds_dynamic commands.<br> ><br> > For reference, the explain plan of get VDS is something like this:<br> ><br> > QUERY PLAN<br> ><br> > ----------------------------------------------------------------------= ---------------------------------------------------------------------------= ----------<br> > Nested Loop (cost=3D9.30..46.75 rows=3D6 width=3D9060) (ac= tual<br> > time=3D0.063..0.068 rows=3D1 loops=3D1)<br> > Join Filter: (vds_static.vds_id =3D vds_statistics.vds_id= )<br> > -> Seq Scan on vds_statistics (cost=3D0.00= ..1.01 rows=3D1 width=3D109)<br> > (actual time=3D0.008..0.008 rows=3D1 loops=3D1)<br> > -> Nested Loop (cost=3D9.30..45.64 rows=3D= 6 width=3D8983) (actual<br> > time=3D0.048..0.052 rows=3D1 loops=3D1)<br> > Join Filter: (vds_groups.vds_group_i= d =3D vds_static.vds_group_id)<br> > -> Nested Loop Left Join &n= bsp;(cost=3D0.00..9.29 rows=3D1 width=3D1389)<br> > (actual time=3D0.013..0.013 rows=3D1 loops=3D1)<br> > -> Seq= Scan on vds_groups (cost=3D0.00..1.01 rows=3D1<br> > width=3D1271) (actual time=3D0.003..0.003 rows=3D1 loops=3D1)<br> > -> Ind= ex Scan using pk_storage_pool on storage_pool<br> > (cost=3D0.00..8.27 rows=3D1 width=3D134) (actual time=3D0.008..0= .008 rows=3D1<br> > loops=3D1)<br> > = Index Cond: (vds_groups.storage_pool_id =3D id)<br> > -> Hash Right Join (c= ost=3D9.30..36.28 rows=3D6 width=3D7610) (actual<br> > time=3D0.033..0.037 rows=3D1 loops=3D1)<br> > Hash Cond: (vds= _spm_id_map.vds_id =3D vds_static.vds_id)<br> > -> Seq= Scan on vds_spm_id_map (cost=3D0.00..22.30 rows=3D1230<br> > width=3D20) (actual time=3D0.003..0.003 rows=3D1 loops=3D1)<br> > -> Has= h (cost=3D9.29..9.29 rows=3D1 width=3D7606) (actual<br> > time=3D0.019..0.019 rows=3D1 loops=3D1)<br> > = Buckets: 1024 Batches: 1 Memory Usage: 2kB<br> > = -> Nested Loop (cost=3D0.00..9.29 rows=3D1 width=3D760= 6)<br> > (actual time=3D0.012..0.013 rows=3D1 loops=3D1)<br> > = -> Seq Scan on vds_dynamic (cost= =3D0.00..1.01<br> > rows=3D1 width=3D1895) (actual time=3D0.006..0.006 rows=3D1 loops=3D1)= <br> > = -> Index Scan using pk_vds_static on vd= s_static<br> > (cost=3D0.00..8.27 rows=3D1 width=3D5711) (actual time=3D0.005..= 0.006 rows=3D1<br> > loops=3D1)<br> > = Index Cond: (vds_id =3D vds= _dynamic.vds_id)<br> > Total runtime: 0.299 ms<br> > (19 rows)<br> ><br> > It's terrible. Adding any additional join will make this worse. Please= <br> > don't add any more tables...<br> <br> </div></div>Thank you for the detailed explanation, my comments:<br> <br> * a very long time isn't an argument for not adding another table (should b= e neglectable);<br> currently we have an unrelated problem, we need to solve it.<br></blockquot= e><div>Of course it is. A very long time for a query that you execute many = times is THE factor. Who said the join has no performance effect? Have you = tested it? Under load? Under many writes/updates? </div> <blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p= x #ccc solid;padding-left:1ex"> <br> * > We start with VDS because in an idle system, with 200 hosts and seve= ral<br> <div class=3D"">> thousands VMs, this is what you get as the top queries= against the<br> > database.<br> <br> </div>so, if fetching VMs takes 10 minutes? and its get called a single tim= e?<br></blockquote><div> Where do you see 10 minutes? If you are looki= ng at the red bar it's the inherent time - total query time * number of que= ries. </div> <blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p= x #ccc solid;padding-left:1ex"> <br> * you didn't reply on my of my suggestion of constructing the VDS records i= n the DB without using joins.<br></blockquote><div>If you mean materialized= views - we don't have it in Postgres just yet... And even if we do, since = we do many updates to vds_statistics and vds_dynamic - I'm not sure it will= have positive impact on our performance. If you mean joins in the database= - everything that is based on VDS is done in the database. Part of the pro= blem, since we can cache some information and only query the dynamic/statis= tics part of VDS, but that's another matter. </div> <blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p= x #ccc solid;padding-left:1ex"> <div class=3D"HOEnZb"><div class=3D"h5"><br> <br> ><br> > ><br> > > > ><br> > > > > ----- Original Message -----<br> > > > > > From: "Gilad Chaplik" <<a href=3D"mailto:gchapl= ik@redhat.com" target=3D"_blank">gchaplik@redhat.com</a>><br> > > > > > To: "Liran Zelkha" <<a href=3D"mailto:liran.zel= kha@gmail.com" target=3D"_blank">liran.zelkha@gmail.com</a>><br> > > > > > Cc: <a href=3D"mailto:devel@linode01.ovirt.org" ta= rget=3D"_blank">devel@linode01.ovirt.org</a>, "engine-devel" <<a href=3D= "mailto:engine-devel@ovirt.org" target=3D"_blank">engine-devel@ovirt.org</a= lit<br> > > > > > > > > > it= to the two existing tables (BTW - relevant for VM<br> > > as<br> > > > > well)<br> > > > > > > > > ><br> > > > > > > > > ><br> > > > > > > > > > but we don't upd= ate it unless the status has changed,<br> > > which<br> > > > > is a<br> > > > > > > > > > rare occurance?<= br> > > > > > > > > ><br> > > > > > > > > > Actually - no. We can definite= ly see times we are updating<br> > > > > > > > > > vds_dynamic<br> > > > > > > > > > with no reason at all. I tried= to create patches for that -<br> > > but<br> > > > > it<br> > > > > > > > > > happens from many different pl= aces in the code.<br> > > > > > > > ><br> > > > > > > > > what would be updated vds_dyanmic f= or status not originating in<br> > > > > update<br> > > > > > > > > run time info?<br> > > > > > > ><br> > > > > > > > We have separate DB flows for that (upda= teStatus and<br> > > > > > > > updatePartialVdsDynamicCalc and more in<= br> > > VdsDynamicDAODbFacadeImpl).<br> > > > > > > > A question: do you know if we update sta= tus in updateVdsDynamic?<br> > > :-)<br> > > > > not<br> > > > > > > > sure but I found a possible race for pen= ding resources (cpu,<br> > > mem),<br> > > > > LOL<br> > > > > > > > :-)<br> > > > > > > ><br> > > > > > > > I think we do but not sure. Will check.<= br> > > > > ><br> > > > > > Of course it is, that was a rhetorical question :-= ) (a lot of<br> > > emoticons<br> > > > > and<br> > > > > > LOLs ;-))<br> > > > > ><br> > > > > > ><br> > > > > > ><br> > > > > > > > Still holds my original thought for havi= ng vds_on_boot.<br> > > > > > > ><br> > > > > > > ><br> > > > > > > ><br> > > > > > > Let's talk f2f on Tuesday?<br> > > > > ><br> > > > > > I'd prefer to reach conclusions here, I'd like eve= ryone to be<br> > > involved<br> > > > > in a<br> > > > > > root issue like this one.<br> > > > ><br> > > ><br> > > > What is the update frequency of this field?<br> > > ><br> > ><br> > > which field?<br> > > status? pending resources? on boot fields?<br> > > iinm, status is updated mostly by user actions, at least in posit= ive<br> > > scenarios, and not that often.<br> > ><br> > ><br> > > > > ><br> > > > > > ><br> > > > > > _______________________________________________<br=
> > > > > Devel mailing list<br> > > > > > <a href=3D"mailto:Devel@ovirt.org" target=3D"_blan= k">Devel@ovirt.org</a><br> > > > > > <a href=3D"http://lists.ovirt.org/mailman/listinfo= /devel" target=3D"_blank">http://lists.ovirt.org/mailman/listinfo/devel</a>= <br> > > > > ><br> > > > ><br> > > ><br> > ><br> ><br> </div></div></blockquote></div><br></div></div> <br>_______________________________________________<br>Devel mailing list<b= r>Devel@ovirt.org<br>http://lists.ovirt.org/mailman/listinfo/devel</blockqu= ote><div><br></div></div></body></html> ------=_Part_3278465_796563103.1397046843672--
On Wed, Apr 9, 2014 at 3:34 PM, Yaniv Dary <ydary@redhat.com> wrote:
Why not move only status with changes a lot to statistics and leave everything as is?
Exactly. No need for a new table. Use the existing ones.
Yaniv
________________________________
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Monday, April 7, 2014 8:51:00 AM
Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 11:03 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 8:51:02 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Kobi Ianko" <kobi@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:40:13 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> wrote:
Joining in... From my point of view, in real life a user should have that many VDSs on one Engine (from a DB point of view). Modern DB system handles tables with millions of records and many relations, Do we really have a performance issue here? We could prefer a more easy to maintain implantation in this case over DB performance
Yes we do. We make many queries on the VDS view, which is a VERY complex view.
Actually I quite agree with Kobi, what is the plan for VMs? why do we start with VDS... what is the biggest deploy do you know of?
We start with VDS because in an idle system, with 200 hosts and several thousands VMs, this is what you get as the top queries against the database. Look at how many times getvds is called. [image: Inline image 1] BTW - the second query is an example of abusing the dynamic query mechanism. The 4th query (an update command) is a set of useless update_vds_dynamic commands.
For reference, the explain plan of get VDS is something like this:
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual time=0.063..0.068 rows=1 loops=1) Join Filter: (vds_static.vds_id = vds_statistics.vds_id) -> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 width=109) (actual time=0.008..0.008 rows=1 loops=1) -> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual time=0.048..0.052 rows=1 loops=1) Join Filter: (vds_groups.vds_group_id = vds_static.vds_group_id) -> Nested Loop Left Join (cost=0.00..9.29 rows=1 width=1389) (actual time=0.013..0.013 rows=1 loops=1) -> Seq Scan on vds_groups (cost=0.00..1.01 rows=1 width=1271) (actual time=0.003..0.003 rows=1 loops=1) -> Index Scan using pk_storage_pool on storage_pool (cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 rows=1 loops=1) Index Cond: (vds_groups.storage_pool_id = id) -> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) (actual time=0.033..0.037 rows=1 loops=1) Hash Cond: (vds_spm_id_map.vds_id = vds_static.vds_id) -> Seq Scan on vds_spm_id_map (cost=0.00..22.30 rows=1230 width=20) (actual time=0.003..0.003 rows=1 loops=1) -> Hash (cost=9.29..9.29 rows=1 width=7606) (actual time=0.019..0.019 rows=1 loops=1) Buckets: 1024 Batches: 1 Memory Usage: 2kB -> Nested Loop (cost=0.00..9.29 rows=1 width=7606) (actual time=0.012..0.013 rows=1 loops=1) -> Seq Scan on vds_dynamic (cost=0.00..1.01 rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1) -> Index Scan using pk_vds_static on vds_static (cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 rows=1 loops=1) Index Cond: (vds_id = vds_dynamic.vds_id) Total runtime: 0.299 ms (19 rows)
It's terrible. Adding any additional join will make this worse. Please don't add any more tables...
Thank you for the detailed explanation, my comments:
* a very long time isn't an argument for not adding another table (should be neglectable); currently we have an unrelated problem, we need to solve it.
Of course it is. A very long time for a query that you execute many times is THE factor. Who said the join has no performance effect? Have you tested it? Under load? Under many writes/updates?
* > We start with VDS because in an idle system, with 200 hosts and several
thousands VMs, this is what you get as the top queries against the database.
so, if fetching VMs takes 10 minutes? and its get called a single time?
Where do you see 10 minutes? If you are looking at the red bar it's the inherent time - total query time * number of queries.
* you didn't reply on my of my suggestion of constructing the VDS records in the DB without using joins.
If you mean materialized views - we don't have it in Postgres just yet... And even if we do, since we do many updates to vds_statistics and vds_dynamic - I'm not sure it will have positive impact on our performance. If you mean joins in the database - everything that is based on VDS is done in the database. Part of the problem, since we can cache some information and only query the dynamic/statistics part of VDS, but that's another matter.
----- Original Message ----- > From: "Gilad Chaplik" <gchaplik@redhat.com> > To: "Liran Zelkha" <liran.zelkha@gmail.com> > Cc: devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org
> Sent: Sunday, April 6, 2014 3:32:26 PM > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > ----- Original Message ----- > > From: "Liran Zelkha" <liran.zelkha@gmail.com> > > To: "Gilad Chaplik" <gchaplik@redhat.com> > > Cc: "Itamar Heim" <iheim@redhat.com>, devel@linode01.ovirt.org, > > "engine-devel" <engine-devel@ovirt.org> > > Sent: Sunday, April 6, 2014 3:26:24 PM > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik <gchaplik@redhat.com
wrote: > > > > > ----- Original Message ----- > > > > From: "Itamar Heim" <iheim@redhat.com> > > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, > > > "engine-devel" <engine-devel@ovirt.org> > > > > Sent: Sunday, April 6, 2014 11:33:12 AM > > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > > > On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > > > > > > > > > > > > > > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim < iheim@redhat.com > > > > > <mailto:iheim@redhat.com>> wrote: > > > > > > > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > > > > > > > > > The problem is with both updates and selects. > > > > > For selects - to get all the information for the VDS we have > > > > > multiple > > > > > joins. Adding another one will hurt performance even more. > > > > > For updates - we have vds_static thats hardly changed. > > > > > vds_statistics > > > > > that changes all the time. vds_dynamic is not changed allot - > > > but > > > > > is > > > > > updated all the time because of the status. I think it's best > > > to > > > > > split > > > > > it to the two existing tables (BTW - relevant for VM as well) > > > > > > > > > > > > > > > but we don't update it unless the status has changed, which is a > > > > > rare occurance? > > > > > > > > > > Actually - no. We can definitely see times we are updating > > > > > vds_dynamic > > > > > with no reason at all. I tried to create patches for that - but it > > > > > happens from many different places in the code. > > > > > > > > what would be updated vds_dyanmic for status not originating in update > > > > run time info? > > > > > > We have separate DB flows for that (updateStatus and > > > updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). > > > A question: do you know if we update status in updateVdsDynamic? :-) not > > > sure but I found a possible race for pending resources (cpu, mem), LOL > > > :-) > > > > > > I think we do but not sure. Will check. > > Of course it is, that was a rhetorical question :-) (a lot of emoticons and > LOLs ;-)) > > > > > > > > Still holds my original thought for having vds_on_boot. > > > > > > > > > > > Let's talk f2f on Tuesday? > > I'd prefer to reach conclusions here, I'd like everyone to be involved in a > root issue like this one.
What is the update frequency of this field?
which field? status? pending resources? on boot fields? iinm, status is updated mostly by user actions, at least in positive scenarios, and not that often.
> > > > _______________________________________________ > Devel mailing list > Devel@ovirt.org > http://lists.ovirt.org/mailman/listinfo/devel >
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Yaniv Dary" <ydary@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Wednesday, April 9, 2014 4:22:39 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Wed, Apr 9, 2014 at 3:34 PM, Yaniv Dary <ydary@redhat.com> wrote:
Why not move only status with changes a lot to statistics and leave everything as is?
Exactly. No need for a new table. Use the existing ones.
Why should the dwh read entire dynamic record for each status change/available resources change? (for VMs as well).
Yaniv
________________________________
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Monday, April 7, 2014 8:51:00 AM
Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 11:03 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 8:51:02 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Kobi Ianko" <kobi@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 3:40:13 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> wrote:
> Joining in... > From my point of view, in real life a user should have that many > VDSs on > one Engine (from a DB point of view). > Modern DB system handles tables with millions of records and many > relations, Do we really have a performance issue here? > We could prefer a more easy to maintain implantation in this case > over DB > performance > > Yes we do. We make many queries on the VDS view, which is a VERY complex view.
Actually I quite agree with Kobi, what is the plan for VMs? why do we start with VDS... what is the biggest deploy do you know of?
We start with VDS because in an idle system, with 200 hosts and several thousands VMs, this is what you get as the top queries against the database. Look at how many times getvds is called. [image: Inline image 1] BTW - the second query is an example of abusing the dynamic query mechanism. The 4th query (an update command) is a set of useless update_vds_dynamic commands.
For reference, the explain plan of get VDS is something like this:
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual time=0.063..0.068 rows=1 loops=1) Join Filter: (vds_static.vds_id = vds_statistics.vds_id) -> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 width=109) (actual time=0.008..0.008 rows=1 loops=1) -> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual time=0.048..0.052 rows=1 loops=1) Join Filter: (vds_groups.vds_group_id = vds_static.vds_group_id) -> Nested Loop Left Join (cost=0.00..9.29 rows=1 width=1389) (actual time=0.013..0.013 rows=1 loops=1) -> Seq Scan on vds_groups (cost=0.00..1.01 rows=1 width=1271) (actual time=0.003..0.003 rows=1 loops=1) -> Index Scan using pk_storage_pool on storage_pool (cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 rows=1 loops=1) Index Cond: (vds_groups.storage_pool_id = id) -> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) (actual time=0.033..0.037 rows=1 loops=1) Hash Cond: (vds_spm_id_map.vds_id = vds_static.vds_id) -> Seq Scan on vds_spm_id_map (cost=0.00..22.30 rows=1230 width=20) (actual time=0.003..0.003 rows=1 loops=1) -> Hash (cost=9.29..9.29 rows=1 width=7606) (actual time=0.019..0.019 rows=1 loops=1) Buckets: 1024 Batches: 1 Memory Usage: 2kB -> Nested Loop (cost=0.00..9.29 rows=1 width=7606) (actual time=0.012..0.013 rows=1 loops=1) -> Seq Scan on vds_dynamic (cost=0.00..1.01 rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1) -> Index Scan using pk_vds_static on vds_static (cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 rows=1 loops=1) Index Cond: (vds_id = vds_dynamic.vds_id) Total runtime: 0.299 ms (19 rows)
It's terrible. Adding any additional join will make this worse. Please don't add any more tables...
Thank you for the detailed explanation, my comments:
* a very long time isn't an argument for not adding another table (should be neglectable); currently we have an unrelated problem, we need to solve it.
Of course it is. A very long time for a query that you execute many times is THE factor. Who said the join has no performance effect? Have you tested it? Under load? Under many writes/updates?
* > We start with VDS because in an idle system, with 200 hosts and several
thousands VMs, this is what you get as the top queries against the database.
so, if fetching VMs takes 10 minutes? and its get called a single time?
Where do you see 10 minutes? If you are looking at the red bar it's the inherent time - total query time * number of queries.
* you didn't reply on my of my suggestion of constructing the VDS records in the DB without using joins.
If you mean materialized views - we don't have it in Postgres just yet... And even if we do, since we do many updates to vds_statistics and vds_dynamic - I'm not sure it will have positive impact on our performance. If you mean joins in the database - everything that is based on VDS is done in the database. Part of the problem, since we can cache some information and only query the dynamic/statistics part of VDS, but that's another matter.
> > ----- Original Message ----- > > From: "Gilad Chaplik" <gchaplik@redhat.com> > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > Cc: devel@linode01.ovirt.org, "engine-devel" > > <engine-devel@ovirt.org
> > Sent: Sunday, April 6, 2014 3:32:26 PM > > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > > > ----- Original Message ----- > > > From: "Liran Zelkha" <liran.zelkha@gmail.com> > > > To: "Gilad Chaplik" <gchaplik@redhat.com> > > > Cc: "Itamar Heim" <iheim@redhat.com>, > > > devel@linode01.ovirt.org, > > > "engine-devel" <engine-devel@ovirt.org> > > > Sent: Sunday, April 6, 2014 3:26:24 PM > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik > > > <gchaplik@redhat.com
> wrote: > > > > > > > ----- Original Message ----- > > > > > From: "Itamar Heim" <iheim@redhat.com> > > > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, > devel@linode01.ovirt.org, > > > > "engine-devel" <engine-devel@ovirt.org> > > > > > Sent: Sunday, April 6, 2014 11:33:12 AM > > > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > > > > > On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > > > > > > > > > > > > > > > > > > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim < iheim@redhat.com > > > > > > <mailto:iheim@redhat.com>> wrote: > > > > > > > > > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > > > > > > > > > > > The problem is with both updates and selects. > > > > > > For selects - to get all the information for the > > > > > > VDS we > have > > > > > > multiple > > > > > > joins. Adding another one will hurt performance > > > > > > even > more. > > > > > > For updates - we have vds_static thats hardly changed. > > > > > > vds_statistics > > > > > > that changes all the time. vds_dynamic is not > > > > > > changed > allot - > > > > but > > > > > > is > > > > > > updated all the time because of the status. I > > > > > > think it's > best > > > > to > > > > > > split > > > > > > it to the two existing tables (BTW - relevant > > > > > > for VM as > well) > > > > > > > > > > > > > > > > > > but we don't update it unless the status has > > > > > > changed, which > is a > > > > > > rare occurance? > > > > > > > > > > > > Actually - no. We can definitely see times we are > > > > > > updating > > > > > > vds_dynamic > > > > > > with no reason at all. I tried to create patches for > > > > > > that - but > it > > > > > > happens from many different places in the code. > > > > > > > > > > what would be updated vds_dyanmic for status not > > > > > originating in > update > > > > > run time info? > > > > > > > > We have separate DB flows for that (updateStatus and > > > > updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). > > > > A question: do you know if we update status in > > > > updateVdsDynamic? :-) > not > > > > sure but I found a possible race for pending resources (cpu, mem), > LOL > > > > :-) > > > > > > > > I think we do but not sure. Will check. > > > > Of course it is, that was a rhetorical question :-) (a lot of emoticons > and > > LOLs ;-)) > > > > > > > > > > > > Still holds my original thought for having vds_on_boot. > > > > > > > > > > > > > > > Let's talk f2f on Tuesday? > > > > I'd prefer to reach conclusions here, I'd like everyone to be involved > in a > > root issue like this one. >
What is the update frequency of this field?
which field? status? pending resources? on boot fields? iinm, status is updated mostly by user actions, at least in positive scenarios, and not that often.
> > > > > > > _______________________________________________ > > Devel mailing list > > Devel@ovirt.org > > http://lists.ovirt.org/mailman/listinfo/devel > > >
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Yaniv Dary" <ydary@redhat.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Thursday, April 10, 2014 11:07:12 AM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Yaniv Dary" <ydary@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Wednesday, April 9, 2014 4:22:39 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Wed, Apr 9, 2014 at 3:34 PM, Yaniv Dary <ydary@redhat.com> wrote:
Why not move only status with changes a lot to statistics and leave everything as is?
Exactly. No need for a new table. Use the existing ones.
Why should the dwh read entire dynamic record for each status change/available resources change? (for VMs as well).
The DWH doesn't work like that. We create views over dynamic\statistics and dynamic\static they go the the configuration and stats tables on the history db side.
From dynamic\statistics we collect every minute no matter what and from dynamic\static we collect only when update_date changes in static. We can not rely on the dynamic update_date since it changes so much, so from my point of view if we can cause minimal changes in dynamic and move status to statistics it would the best thing. What we have currently is a task that run once a hour and checks if any change was made in the dynamic table (via very ugly joins and rejects) and sync is there was any change. It would even be better if you also move swap_size which doesn't change much to dynamic.
Yaniv
________________________________
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Monday, April 7, 2014 8:51:00 AM
Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 11:03 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 8:51:02 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message ----- > From: "Liran Zelkha" <liran.zelkha@gmail.com> > To: "Kobi Ianko" <kobi@redhat.com> > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, > devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> > Sent: Sunday, April 6, 2014 3:40:13 PM > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> > wrote: > > > Joining in... > > From my point of view, in real life a user should have that many > > VDSs on > > one Engine (from a DB point of view). > > Modern DB system handles tables with millions of records and > > many > > relations, Do we really have a performance issue here? > > We could prefer a more easy to maintain implantation in this > > case > > over DB > > performance > > > > Yes we do. We make many queries on the VDS view, which is a VERY complex > view. >
Actually I quite agree with Kobi, what is the plan for VMs? why do we start with VDS... what is the biggest deploy do you know of?
We start with VDS because in an idle system, with 200 hosts and several thousands VMs, this is what you get as the top queries against the database. Look at how many times getvds is called. [image: Inline image 1] BTW - the second query is an example of abusing the dynamic query mechanism. The 4th query (an update command) is a set of useless update_vds_dynamic commands.
For reference, the explain plan of get VDS is something like this:
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual time=0.063..0.068 rows=1 loops=1) Join Filter: (vds_static.vds_id = vds_statistics.vds_id) -> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 width=109) (actual time=0.008..0.008 rows=1 loops=1) -> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual time=0.048..0.052 rows=1 loops=1) Join Filter: (vds_groups.vds_group_id = vds_static.vds_group_id) -> Nested Loop Left Join (cost=0.00..9.29 rows=1 width=1389) (actual time=0.013..0.013 rows=1 loops=1) -> Seq Scan on vds_groups (cost=0.00..1.01 rows=1 width=1271) (actual time=0.003..0.003 rows=1 loops=1) -> Index Scan using pk_storage_pool on storage_pool (cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 rows=1 loops=1) Index Cond: (vds_groups.storage_pool_id = id) -> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) (actual time=0.033..0.037 rows=1 loops=1) Hash Cond: (vds_spm_id_map.vds_id = vds_static.vds_id) -> Seq Scan on vds_spm_id_map (cost=0.00..22.30 rows=1230 width=20) (actual time=0.003..0.003 rows=1 loops=1) -> Hash (cost=9.29..9.29 rows=1 width=7606) (actual time=0.019..0.019 rows=1 loops=1) Buckets: 1024 Batches: 1 Memory Usage: 2kB -> Nested Loop (cost=0.00..9.29 rows=1 width=7606) (actual time=0.012..0.013 rows=1 loops=1) -> Seq Scan on vds_dynamic (cost=0.00..1.01 rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1) -> Index Scan using pk_vds_static on vds_static (cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 rows=1 loops=1) Index Cond: (vds_id = vds_dynamic.vds_id) Total runtime: 0.299 ms (19 rows)
It's terrible. Adding any additional join will make this worse. Please don't add any more tables...
Thank you for the detailed explanation, my comments:
* a very long time isn't an argument for not adding another table (should be neglectable); currently we have an unrelated problem, we need to solve it.
Of course it is. A very long time for a query that you execute many times is THE factor. Who said the join has no performance effect? Have you tested it? Under load? Under many writes/updates?
* > We start with VDS because in an idle system, with 200 hosts and several
thousands VMs, this is what you get as the top queries against the database.
so, if fetching VMs takes 10 minutes? and its get called a single time?
Where do you see 10 minutes? If you are looking at the red bar it's the inherent time - total query time * number of queries.
* you didn't reply on my of my suggestion of constructing the VDS records in the DB without using joins.
If you mean materialized views - we don't have it in Postgres just yet... And even if we do, since we do many updates to vds_statistics and vds_dynamic - I'm not sure it will have positive impact on our performance. If you mean joins in the database - everything that is based on VDS is done in the database. Part of the problem, since we can cache some information and only query the dynamic/statistics part of VDS, but that's another matter.
> > > > ----- Original Message ----- > > > From: "Gilad Chaplik" <gchaplik@redhat.com> > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > Cc: devel@linode01.ovirt.org, "engine-devel" > > > <engine-devel@ovirt.org > > > > Sent: Sunday, April 6, 2014 3:32:26 PM > > > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > > > > > ----- Original Message ----- > > > > From: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > To: "Gilad Chaplik" <gchaplik@redhat.com> > > > > Cc: "Itamar Heim" <iheim@redhat.com>, > > > > devel@linode01.ovirt.org, > > > > "engine-devel" <engine-devel@ovirt.org> > > > > Sent: Sunday, April 6, 2014 3:26:24 PM > > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > > > On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik > > > > <gchaplik@redhat.com > > > wrote: > > > > > > > > > ----- Original Message ----- > > > > > > From: "Itamar Heim" <iheim@redhat.com> > > > > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, > > devel@linode01.ovirt.org, > > > > > "engine-devel" <engine-devel@ovirt.org> > > > > > > Sent: Sunday, April 6, 2014 11:33:12 AM > > > > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > > > > > > > On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim < iheim@redhat.com > > > > > > > <mailto:iheim@redhat.com>> wrote: > > > > > > > > > > > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > > > > > > > > > > > > > The problem is with both updates and selects. > > > > > > > For selects - to get all the information for > > > > > > > the > > > > > > > VDS we > > have > > > > > > > multiple > > > > > > > joins. Adding another one will hurt > > > > > > > performance > > > > > > > even > > more. > > > > > > > For updates - we have vds_static thats hardly changed. > > > > > > > vds_statistics > > > > > > > that changes all the time. vds_dynamic is not > > > > > > > changed > > allot - > > > > > but > > > > > > > is > > > > > > > updated all the time because of the status. I > > > > > > > think it's > > best > > > > > to > > > > > > > split > > > > > > > it to the two existing tables (BTW - relevant > > > > > > > for VM as > > well) > > > > > > > > > > > > > > > > > > > > > but we don't update it unless the status has > > > > > > > changed, which > > is a > > > > > > > rare occurance? > > > > > > > > > > > > > > Actually - no. We can definitely see times we are > > > > > > > updating > > > > > > > vds_dynamic > > > > > > > with no reason at all. I tried to create patches for > > > > > > > that - but > > it > > > > > > > happens from many different places in the code. > > > > > > > > > > > > what would be updated vds_dyanmic for status not > > > > > > originating in > > update > > > > > > run time info? > > > > > > > > > > We have separate DB flows for that (updateStatus and > > > > > updatePartialVdsDynamicCalc and more in VdsDynamicDAODbFacadeImpl). > > > > > A question: do you know if we update status in > > > > > updateVdsDynamic? :-) > > not > > > > > sure but I found a possible race for pending resources > > > > > (cpu, mem), > > LOL > > > > > :-) > > > > > > > > > > I think we do but not sure. Will check. > > > > > > Of course it is, that was a rhetorical question :-) (a lot of emoticons > > and > > > LOLs ;-)) > > > > > > > > > > > > > > > > Still holds my original thought for having vds_on_boot. > > > > > > > > > > > > > > > > > > > Let's talk f2f on Tuesday? > > > > > > I'd prefer to reach conclusions here, I'd like everyone to be involved > > in a > > > root issue like this one. > > > > What is the update frequency of this field? >
which field? status? pending resources? on boot fields? iinm, status is updated mostly by user actions, at least in positive scenarios, and not that often.
> > > > > > > > > > _______________________________________________ > > > Devel mailing list > > > Devel@ovirt.org > > > http://lists.ovirt.org/mailman/listinfo/devel > > > > > >
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
----- Original Message -----
From: "Yaniv Dary" <ydary@redhat.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Liran Zelkha" <liran.zelkha@gmail.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Thursday, April 10, 2014 11:44:32 AM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Yaniv Dary" <ydary@redhat.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Thursday, April 10, 2014 11:07:12 AM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Yaniv Dary" <ydary@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Wednesday, April 9, 2014 4:22:39 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Wed, Apr 9, 2014 at 3:34 PM, Yaniv Dary <ydary@redhat.com> wrote:
Why not move only status with changes a lot to statistics and leave everything as is?
Exactly. No need for a new table. Use the existing ones.
Why should the dwh read entire dynamic record for each status change/available resources change? (for VMs as well).
The DWH doesn't work like that. We create views over dynamic\statistics and dynamic\static they go the the configuration and stats tables on the history db side. From dynamic\statistics we collect every minute no matter what and from dynamic\static we collect only when update_date changes in static.
dynamic\statistics - dynamic\static to fully understand it, we have 2 triggers to collect from dynamic?
We can not rely on the dynamic update_date since it changes so much, so from my point of view if we can cause minimal changes in dynamic and move status to statistics it would the best thing. What we have currently is a task that run once a hour and checks if any change was made in the dynamic table (via very ugly joins and rejects) and sync is there was any change. It would even be better if you also move swap_size which doesn't change much to dynamic.
oh... I thought that we had a trigger to dynamic (if ^ is correct we have). +1 (out of 1) for you comment :-) and thanks Yaniv the elaborate explanation!
Yaniv
________________________________
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Monday, April 7, 2014 8:51:00 AM
Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 11:03 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Sunday, April 6, 2014 8:51:02 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
> ----- Original Message ----- > > From: "Liran Zelkha" <liran.zelkha@gmail.com> > > To: "Kobi Ianko" <kobi@redhat.com> > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, > > devel@linode01.ovirt.org, > "engine-devel" <engine-devel@ovirt.org> > > Sent: Sunday, April 6, 2014 3:40:13 PM > > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > > > On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> > > wrote: > > > > > Joining in... > > > From my point of view, in real life a user should have that > > > many > > > VDSs > on > > > one Engine (from a DB point of view). > > > Modern DB system handles tables with millions of records and > > > many > > > relations, Do we really have a performance issue here? > > > We could prefer a more easy to maintain implantation in this > > > case > > > over > DB > > > performance > > > > > > Yes we do. We make many queries on the VDS view, which is a > > > VERY > complex > > view. > > > > Actually I quite agree with Kobi, what is the plan for VMs? why do > we > start with VDS... > what is the biggest deploy do you know of? > We start with VDS because in an idle system, with 200 hosts and several thousands VMs, this is what you get as the top queries against the database. Look at how many times getvds is called. [image: Inline image 1] BTW - the second query is an example of abusing the dynamic query mechanism. The 4th query (an update command) is a set of useless update_vds_dynamic commands.
For reference, the explain plan of get VDS is something like this:
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------------------------------- Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual time=0.063..0.068 rows=1 loops=1) Join Filter: (vds_static.vds_id = vds_statistics.vds_id) -> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 width=109) (actual time=0.008..0.008 rows=1 loops=1) -> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual time=0.048..0.052 rows=1 loops=1) Join Filter: (vds_groups.vds_group_id = vds_static.vds_group_id) -> Nested Loop Left Join (cost=0.00..9.29 rows=1 width=1389) (actual time=0.013..0.013 rows=1 loops=1) -> Seq Scan on vds_groups (cost=0.00..1.01 rows=1 width=1271) (actual time=0.003..0.003 rows=1 loops=1) -> Index Scan using pk_storage_pool on storage_pool (cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 rows=1 loops=1) Index Cond: (vds_groups.storage_pool_id = id) -> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) (actual time=0.033..0.037 rows=1 loops=1) Hash Cond: (vds_spm_id_map.vds_id = vds_static.vds_id) -> Seq Scan on vds_spm_id_map (cost=0.00..22.30 rows=1230 width=20) (actual time=0.003..0.003 rows=1 loops=1) -> Hash (cost=9.29..9.29 rows=1 width=7606) (actual time=0.019..0.019 rows=1 loops=1) Buckets: 1024 Batches: 1 Memory Usage: 2kB -> Nested Loop (cost=0.00..9.29 rows=1 width=7606) (actual time=0.012..0.013 rows=1 loops=1) -> Seq Scan on vds_dynamic (cost=0.00..1.01 rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1) -> Index Scan using pk_vds_static on vds_static (cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 rows=1 loops=1) Index Cond: (vds_id = vds_dynamic.vds_id) Total runtime: 0.299 ms (19 rows)
It's terrible. Adding any additional join will make this worse. Please don't add any more tables...
Thank you for the detailed explanation, my comments:
* a very long time isn't an argument for not adding another table (should be neglectable); currently we have an unrelated problem, we need to solve it.
Of course it is. A very long time for a query that you execute many times is THE factor. Who said the join has no performance effect? Have you tested it? Under load? Under many writes/updates?
* > We start with VDS because in an idle system, with 200 hosts and several
thousands VMs, this is what you get as the top queries against the database.
so, if fetching VMs takes 10 minutes? and its get called a single time?
Where do you see 10 minutes? If you are looking at the red bar it's the inherent time - total query time * number of queries.
* you didn't reply on my of my suggestion of constructing the VDS records in the DB without using joins.
If you mean materialized views - we don't have it in Postgres just yet... And even if we do, since we do many updates to vds_statistics and vds_dynamic - I'm not sure it will have positive impact on our performance. If you mean joins in the database - everything that is based on VDS is done in the database. Part of the problem, since we can cache some information and only query the dynamic/statistics part of VDS, but that's another matter.
> > > > > > > ----- Original Message ----- > > > > From: "Gilad Chaplik" <gchaplik@redhat.com> > > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > Cc: devel@linode01.ovirt.org, "engine-devel" > > > > <engine-devel@ovirt.org > > > > > > Sent: Sunday, April 6, 2014 3:32:26 PM > > > > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > > > > > > > ----- Original Message ----- > > > > > From: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > > To: "Gilad Chaplik" <gchaplik@redhat.com> > > > > > Cc: "Itamar Heim" <iheim@redhat.com>, > > > > > devel@linode01.ovirt.org, > > > > > "engine-devel" <engine-devel@ovirt.org> > > > > > Sent: Sunday, April 6, 2014 3:26:24 PM > > > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > > > > > On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik > > > > > <gchaplik@redhat.com > > > > > wrote: > > > > > > > > > > > ----- Original Message ----- > > > > > > > From: "Itamar Heim" <iheim@redhat.com> > > > > > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > > > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, > > > devel@linode01.ovirt.org, > > > > > > "engine-devel" <engine-devel@ovirt.org> > > > > > > > Sent: Sunday, April 6, 2014 11:33:12 AM > > > > > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > > > > > > > > > On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim < > iheim@redhat.com > > > > > > > > <mailto:iheim@redhat.com>> wrote: > > > > > > > > > > > > > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > > > > > > > > > > > > > > > The problem is with both updates and > > > > > > > > selects. > > > > > > > > For selects - to get all the information for > > > > > > > > the > > > > > > > > VDS > we > > > have > > > > > > > > multiple > > > > > > > > joins. Adding another one will hurt > > > > > > > > performance > > > > > > > > even > > > more. > > > > > > > > For updates - we have vds_static thats > > > > > > > > hardly > changed. > > > > > > > > vds_statistics > > > > > > > > that changes all the time. vds_dynamic is > > > > > > > > not > > > > > > > > changed > > > allot - > > > > > > but > > > > > > > > is > > > > > > > > updated all the time because of the status. > > > > > > > > I > > > > > > > > think > it's > > > best > > > > > > to > > > > > > > > split > > > > > > > > it to the two existing tables (BTW - > > > > > > > > relevant > > > > > > > > for VM > as > > > well) > > > > > > > > > > > > > > > > > > > > > > > > but we don't update it unless the status has > > > > > > > > changed, > which > > > is a > > > > > > > > rare occurance? > > > > > > > > > > > > > > > > Actually - no. We can definitely see times we are > > > > > > > > updating > > > > > > > > vds_dynamic > > > > > > > > with no reason at all. I tried to create patches for > > > > > > > > that - > but > > > it > > > > > > > > happens from many different places in the code. > > > > > > > > > > > > > > what would be updated vds_dyanmic for status not > > > > > > > originating in > > > update > > > > > > > run time info? > > > > > > > > > > > > We have separate DB flows for that (updateStatus and > > > > > > updatePartialVdsDynamicCalc and more in > VdsDynamicDAODbFacadeImpl). > > > > > > A question: do you know if we update status in > > > > > > updateVdsDynamic? > :-) > > > not > > > > > > sure but I found a possible race for pending resources > > > > > > (cpu, > mem), > > > LOL > > > > > > :-) > > > > > > > > > > > > I think we do but not sure. Will check. > > > > > > > > Of course it is, that was a rhetorical question :-) (a lot > > > > of > emoticons > > > and > > > > LOLs ;-)) > > > > > > > > > > > > > > > > > > > > Still holds my original thought for having vds_on_boot. > > > > > > > > > > > > > > > > > > > > > > > Let's talk f2f on Tuesday? > > > > > > > > I'd prefer to reach conclusions here, I'd like everyone to > > > > be > involved > > > in a > > > > root issue like this one. > > > > > > > What is the update frequency of this field? > > > > which field? > status? pending resources? on boot fields? > iinm, status is updated mostly by user actions, at least in > positive > scenarios, and not that often. > > > > > > > > > > > > > > > _______________________________________________ > > > > Devel mailing list > > > > Devel@ovirt.org > > > > http://lists.ovirt.org/mailman/listinfo/devel > > > > > > > > > >
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Yaniv Dary" <ydary@redhat.com> Cc: "Liran Zelkha" <liran.zelkha@gmail.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Thursday, April 10, 2014 12:30:52 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Yaniv Dary" <ydary@redhat.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Liran Zelkha" <liran.zelkha@gmail.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Thursday, April 10, 2014 11:44:32 AM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Gilad Chaplik" <gchaplik@redhat.com> To: "Liran Zelkha" <liran.zelkha@gmail.com> Cc: "Yaniv Dary" <ydary@redhat.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Thursday, April 10, 2014 11:07:12 AM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
----- Original Message -----
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Yaniv Dary" <ydary@redhat.com> Cc: "Gilad Chaplik" <gchaplik@redhat.com>, "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Wednesday, April 9, 2014 4:22:39 PM Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Wed, Apr 9, 2014 at 3:34 PM, Yaniv Dary <ydary@redhat.com> wrote:
Why not move only status with changes a lot to statistics and leave everything as is?
Exactly. No need for a new table. Use the existing ones.
Why should the dwh read entire dynamic record for each status change/available resources change? (for VMs as well).
The DWH doesn't work like that. We create views over dynamic\statistics and dynamic\static they go the the configuration and stats tables on the history db side. From dynamic\statistics we collect every minute no matter what and from dynamic\static we collect only when update_date changes in static.
dynamic\statistics - dynamic\static
to fully understand it, we have 2 triggers to collect from dynamic?
No, the correct answer is that we don't have any triggers. dynamic\statistics - we collect every minute for stats like status and cpu usage no matter what changed. dynamic\static - we collect either when static table changes and update_date change or once an hour if dynamic changes, but we check this with join and rejects. This is because dynamic changes too much right now.
We can not rely on the dynamic update_date since it changes so much, so from my point of view if we can cause minimal changes in dynamic and move status to statistics it would the best thing. What we have currently is a task that run once a hour and checks if any change was made in the dynamic table (via very ugly joins and rejects) and sync is there was any change. It would even be better if you also move swap_size which doesn't change much to dynamic.
oh... I thought that we had a trigger to dynamic (if ^ is correct we have).
+1 (out of 1) for you comment :-) and thanks Yaniv the elaborate explanation!
Yaniv
________________________________
From: "Liran Zelkha" <liran.zelkha@gmail.com> To: "Gilad Chaplik" <gchaplik@redhat.com> Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, "engine-devel" <engine-devel@ovirt.org> Sent: Monday, April 7, 2014 8:51:00 AM
Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor
On Sun, Apr 6, 2014 at 11:03 PM, Gilad Chaplik <gchaplik@redhat.com> wrote:
----- Original Message ----- > From: "Liran Zelkha" <liran.zelkha@gmail.com> > To: "Gilad Chaplik" <gchaplik@redhat.com> > Cc: "Kobi Ianko" <kobi@redhat.com>, devel@linode01.ovirt.org, > "engine-devel" <engine-devel@ovirt.org> > Sent: Sunday, April 6, 2014 8:51:02 PM > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > On Sun, Apr 6, 2014 at 6:32 PM, Gilad Chaplik > <gchaplik@redhat.com> > wrote: > > > ----- Original Message ----- > > > From: "Liran Zelkha" <liran.zelkha@gmail.com> > > > To: "Kobi Ianko" <kobi@redhat.com> > > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, > > > devel@linode01.ovirt.org, > > "engine-devel" <engine-devel@ovirt.org> > > > Sent: Sunday, April 6, 2014 3:40:13 PM > > > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > > > > > On Sun, Apr 6, 2014 at 3:37 PM, Kobi Ianko <kobi@redhat.com> > > > wrote: > > > > > > > Joining in... > > > > From my point of view, in real life a user should have that > > > > many > > > > VDSs > > on > > > > one Engine (from a DB point of view). > > > > Modern DB system handles tables with millions of records and > > > > many > > > > relations, Do we really have a performance issue here? > > > > We could prefer a more easy to maintain implantation in this > > > > case > > > > over > > DB > > > > performance > > > > > > > > Yes we do. We make many queries on the VDS view, which is a > > > > VERY > > complex > > > view. > > > > > > > Actually I quite agree with Kobi, what is the plan for VMs? why > > do > > we > > start with VDS... > > what is the biggest deploy do you know of? > > > We start with VDS because in an idle system, with 200 hosts and > several > thousands VMs, this is what you get as the top queries against the > database. Look at how many times getvds is called. > [image: Inline image 1] > BTW - the second query is an example of abusing the dynamic query > mechanism. The 4th query (an update command) is a set of useless > update_vds_dynamic commands. > > For reference, the explain plan of get VDS is something like this: > > QUERY PLAN > > ----------------------------------------------------------------------------------------------------------------------------------------------------------- > Nested Loop (cost=9.30..46.75 rows=6 width=9060) (actual > time=0.063..0.068 rows=1 loops=1) > Join Filter: (vds_static.vds_id = vds_statistics.vds_id) > -> Seq Scan on vds_statistics (cost=0.00..1.01 rows=1 > width=109) > (actual time=0.008..0.008 rows=1 loops=1) > -> Nested Loop (cost=9.30..45.64 rows=6 width=8983) (actual > time=0.048..0.052 rows=1 loops=1) > Join Filter: (vds_groups.vds_group_id = > vds_static.vds_group_id) > -> Nested Loop Left Join (cost=0.00..9.29 rows=1 > width=1389) > (actual time=0.013..0.013 rows=1 loops=1) > -> Seq Scan on vds_groups (cost=0.00..1.01 rows=1 > width=1271) (actual time=0.003..0.003 rows=1 loops=1) > -> Index Scan using pk_storage_pool on > storage_pool > (cost=0.00..8.27 rows=1 width=134) (actual time=0.008..0.008 > rows=1 > loops=1) > Index Cond: (vds_groups.storage_pool_id = id) > -> Hash Right Join (cost=9.30..36.28 rows=6 width=7610) > (actual > time=0.033..0.037 rows=1 loops=1) > Hash Cond: (vds_spm_id_map.vds_id = > vds_static.vds_id) > -> Seq Scan on vds_spm_id_map (cost=0.00..22.30 > rows=1230 > width=20) (actual time=0.003..0.003 rows=1 loops=1) > -> Hash (cost=9.29..9.29 rows=1 width=7606) > (actual > time=0.019..0.019 rows=1 loops=1) > Buckets: 1024 Batches: 1 Memory Usage: 2kB > -> Nested Loop (cost=0.00..9.29 rows=1 > width=7606) > (actual time=0.012..0.013 rows=1 loops=1) > -> Seq Scan on vds_dynamic > (cost=0.00..1.01 > rows=1 width=1895) (actual time=0.006..0.006 rows=1 loops=1) > -> Index Scan using pk_vds_static on > vds_static > (cost=0.00..8.27 rows=1 width=5711) (actual time=0.005..0.006 > rows=1 > loops=1) > Index Cond: (vds_id = > vds_dynamic.vds_id) > Total runtime: 0.299 ms > (19 rows) > > It's terrible. Adding any additional join will make this worse. > Please > don't add any more tables...
Thank you for the detailed explanation, my comments:
* a very long time isn't an argument for not adding another table (should be neglectable); currently we have an unrelated problem, we need to solve it.
Of course it is. A very long time for a query that you execute many times is THE factor. Who said the join has no performance effect? Have you tested it? Under load? Under many writes/updates?
* > We start with VDS because in an idle system, with 200 hosts and several > thousands VMs, this is what you get as the top queries against the > database.
so, if fetching VMs takes 10 minutes? and its get called a single time?
Where do you see 10 minutes? If you are looking at the red bar it's the inherent time - total query time * number of queries.
* you didn't reply on my of my suggestion of constructing the VDS records in the DB without using joins.
If you mean materialized views - we don't have it in Postgres just yet... And even if we do, since we do many updates to vds_statistics and vds_dynamic - I'm not sure it will have positive impact on our performance. If you mean joins in the database - everything that is based on VDS is done in the database. Part of the problem, since we can cache some information and only query the dynamic/statistics part of VDS, but that's another matter.
> > > > > > > > > > > ----- Original Message ----- > > > > > From: "Gilad Chaplik" <gchaplik@redhat.com> > > > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > > Cc: devel@linode01.ovirt.org, "engine-devel" > > > > > <engine-devel@ovirt.org > > > > > > > > Sent: Sunday, April 6, 2014 3:32:26 PM > > > > > Subject: Re: [Devel] [Engine-devel] vds_dynamic refactor > > > > > > > > > > ----- Original Message ----- > > > > > > From: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > > > To: "Gilad Chaplik" <gchaplik@redhat.com> > > > > > > Cc: "Itamar Heim" <iheim@redhat.com>, > > > > > > devel@linode01.ovirt.org, > > > > > > "engine-devel" <engine-devel@ovirt.org> > > > > > > Sent: Sunday, April 6, 2014 3:26:24 PM > > > > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > > > > > > > On Sun, Apr 6, 2014 at 3:18 PM, Gilad Chaplik > > > > > > <gchaplik@redhat.com > > > > > > > wrote: > > > > > > > > > > > > > ----- Original Message ----- > > > > > > > > From: "Itamar Heim" <iheim@redhat.com> > > > > > > > > To: "Liran Zelkha" <liran.zelkha@gmail.com> > > > > > > > > Cc: "Gilad Chaplik" <gchaplik@redhat.com>, > > > > devel@linode01.ovirt.org, > > > > > > > "engine-devel" <engine-devel@ovirt.org> > > > > > > > > Sent: Sunday, April 6, 2014 11:33:12 AM > > > > > > > > Subject: Re: [Engine-devel] vds_dynamic refactor > > > > > > > > > > > > > > > > On 04/06/2014 11:32 AM, Liran Zelkha wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Sun, Apr 6, 2014 at 11:22 AM, Itamar Heim < > > iheim@redhat.com > > > > > > > > > <mailto:iheim@redhat.com>> wrote: > > > > > > > > > > > > > > > > > > On 04/03/2014 07:51 PM, Liran Zelkha wrote: > > > > > > > > > > > > > > > > > > The problem is with both updates and > > > > > > > > > selects. > > > > > > > > > For selects - to get all the information > > > > > > > > > for > > > > > > > > > the > > > > > > > > > VDS > > we > > > > have > > > > > > > > > multiple > > > > > > > > > joins. Adding another one will hurt > > > > > > > > > performance > > > > > > > > > even > > > > more. > > > > > > > > > For updates - we have vds_static thats > > > > > > > > > hardly > > changed. > > > > > > > > > vds_statistics > > > > > > > > > that changes all the time. vds_dynamic is > > > > > > > > > not > > > > > > > > > changed > > > > allot - > > > > > > > but > > > > > > > > > is > > > > > > > > > updated all the time because of the > > > > > > > > > status. > > > > > > > > > I > > > > > > > > > think > > it's > > > > best > > > > > > > to > > > > > > > > > split > > > > > > > > > it to the two existing tables (BTW - > > > > > > > > > relevant > > > > > > > > > for VM > > as > > > > well) > > > > > > > > > > > > > > > > > > > > > > > > > > > but we don't update it unless the status has > > > > > > > > > changed, > > which > > > > is a > > > > > > > > > rare occurance? > > > > > > > > > > > > > > > > > > Actually - no. We can definitely see times we are > > > > > > > > > updating > > > > > > > > > vds_dynamic > > > > > > > > > with no reason at all. I tried to create patches > > > > > > > > > for > > > > > > > > > that - > > but > > > > it > > > > > > > > > happens from many different places in the code. > > > > > > > > > > > > > > > > what would be updated vds_dyanmic for status not > > > > > > > > originating in > > > > update > > > > > > > > run time info? > > > > > > > > > > > > > > We have separate DB flows for that (updateStatus and > > > > > > > updatePartialVdsDynamicCalc and more in > > VdsDynamicDAODbFacadeImpl). > > > > > > > A question: do you know if we update status in > > > > > > > updateVdsDynamic? > > :-) > > > > not > > > > > > > sure but I found a possible race for pending resources > > > > > > > (cpu, > > mem), > > > > LOL > > > > > > > :-) > > > > > > > > > > > > > > I think we do but not sure. Will check. > > > > > > > > > > Of course it is, that was a rhetorical question :-) (a lot > > > > > of > > emoticons > > > > and > > > > > LOLs ;-)) > > > > > > > > > > > > > > > > > > > > > > > > Still holds my original thought for having > > > > > > > vds_on_boot. > > > > > > > > > > > > > > > > > > > > > > > > > > > Let's talk f2f on Tuesday? > > > > > > > > > > I'd prefer to reach conclusions here, I'd like everyone to > > > > > be > > involved > > > > in a > > > > > root issue like this one. > > > > > > > > > > What is the update frequency of this field? > > > > > > > which field? > > status? pending resources? on boot fields? > > iinm, status is updated mostly by user actions, at least in > > positive > > scenarios, and not that often. > > > > > > > > > > > > > > > > > > > > _______________________________________________ > > > > > Devel mailing list > > > > > Devel@ovirt.org > > > > > http://lists.ovirt.org/mailman/listinfo/devel > > > > > > > > > > > > > > >
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
participants (5)
-
Gilad Chaplik -
Itamar Heim -
Kobi Ianko -
Liran Zelkha -
Yaniv Dary