[Users] Fault tolerant storage
Hi, would it be possible in near future to use for example 2 attached DATA storage (iSCSi, or so) as fault tolerant storage? I mean I have two data storage connected to data center and they include exactly same data and my VMs runs from one and all changes are mirrored to second like RAID 1 and if one storage fails everything switch to second storage without user notice?
----- Original Message -----
From: "Jakub Bittner" <j.bittner@nbu.cz> To: users@ovirt.org Sent: Tuesday, April 30, 2013 5:19:12 PM Subject: [Users] Fault tolerant storage
Hi,
would it be possible in near future to use for example 2 attached DATA storage (iSCSi, or so) as fault tolerant storage?
I mean I have two data storage connected to data center and they include exactly same data and my VMs runs from one and all changes are mirrored to second like RAID 1 and if one storage fails everything switch to second storage without user notice?
Hi Jakub, If what you mean is to handle the whole physical storage devices redundancy on your own, and assign one storage domain to this constellation, then this is already possible. It doesn't matter what lies underneath the virtual storage domain. If what you mean is assigning two different Storage domains to two different storage devices then no - there are currently no plans to support this. In case this is the latter and you're going to pursue the issue on your own, we'd appreciate it if you do it upstream. Best Regards, Vered
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
I would be surprised if you could do that without your users noticing. It seems like you would be better served to have each storage domain on a RAID array and rely on the hardware tools of the RAID to handle disk failure tolerance. Cheers, jonathan -----Original Message----- From: users-bounces@ovirt.org [mailto:users-bounces@ovirt.org] On Behalf Of Jakub Bittner Sent: Tuesday, April 30, 2013 9:19 AM To: users@ovirt.org Subject: [Users] Fault tolerant storage Hi, would it be possible in near future to use for example 2 attached DATA storage (iSCSi, or so) as fault tolerant storage? I mean I have two data storage connected to data center and they include exactly same data and my VMs runs from one and all changes are mirrored to second like RAID 1 and if one storage fails everything switch to second storage without user notice? _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users ________________________________ This is a PRIVATE message. If you are not the intended recipient, please delete without copying and kindly advise us by e-mail of the mistake in delivery. NOTE: Regardless of content, this e-mail shall not operate to bind SKOPOS to any order or other contract unless pursuant to explicit written agreement or government initiative expressly permitting the use of e-mail for such purpose.
Hi Jakub, I assume the solution you are searching for is something GlusterFS can serve for example: http://www.gluster.org/about/ Best, Sven. -----Ursprüngliche Nachricht----- Von: users-bounces@ovirt.org [mailto:users-bounces@ovirt.org] Im Auftrag von Jonathan Horne Gesendet: Mittwoch, 1. Mai 2013 17:16 An: Jakub Bittner; users@ovirt.org Betreff: Re: [Users] Fault tolerant storage I would be surprised if you could do that without your users noticing. It seems like you would be better served to have each storage domain on a RAID array and rely on the hardware tools of the RAID to handle disk failure tolerance. Cheers, jonathan -----Original Message----- From: users-bounces@ovirt.org [mailto:users-bounces@ovirt.org] On Behalf Of Jakub Bittner Sent: Tuesday, April 30, 2013 9:19 AM To: users@ovirt.org Subject: [Users] Fault tolerant storage Hi, would it be possible in near future to use for example 2 attached DATA storage (iSCSi, or so) as fault tolerant storage? I mean I have two data storage connected to data center and they include exactly same data and my VMs runs from one and all changes are mirrored to second like RAID 1 and if one storage fails everything switch to second storage without user notice? _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users ________________________________ This is a PRIVATE message. If you are not the intended recipient, please delete without copying and kindly advise us by e-mail of the mistake in delivery. NOTE: Regardless of content, this e-mail shall not operate to bind SKOPOS to any order or other contract unless pursuant to explicit written agreement or government initiative expressly permitting the use of e-mail for such purpose. _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Dne 1.5.2013 17:25, Sven Knohsalla napsal(a):
Hi Jakub,
I assume the solution you are searching for is something GlusterFS can serve for example: http://www.gluster.org/about/
Best, Sven.
-----Ursprüngliche Nachricht----- Von: users-bounces@ovirt.org [mailto:users-bounces@ovirt.org] Im Auftrag von Jonathan Horne Gesendet: Mittwoch, 1. Mai 2013 17:16 An: Jakub Bittner; users@ovirt.org Betreff: Re: [Users] Fault tolerant storage
I would be surprised if you could do that without your users noticing.
It seems like you would be better served to have each storage domain on a RAID array and rely on the hardware tools of the RAID to handle disk failure tolerance.
Cheers, jonathan
-----Original Message----- From: users-bounces@ovirt.org [mailto:users-bounces@ovirt.org] On Behalf Of Jakub Bittner Sent: Tuesday, April 30, 2013 9:19 AM To: users@ovirt.org Subject: [Users] Fault tolerant storage
Hi,
would it be possible in near future to use for example 2 attached DATA storage (iSCSi, or so) as fault tolerant storage?
I mean I have two data storage connected to data center and they include exactly same data and my VMs runs from one and all changes are mirrored to second like RAID 1 and if one storage fails everything switch to second storage without user notice? _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
________________________________ This is a PRIVATE message. If you are not the intended recipient, please delete without copying and kindly advise us by e-mail of the mistake in delivery. NOTE: Regardless of content, this e-mail shall not operate to bind SKOPOS to any order or other contract unless pursuant to explicit written agreement or government initiative expressly permitting the use of e-mail for such purpose. _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users Hi,
thank you for replies. Sven: I know Gluster. I am using it since 2009 in production but the problem is that many organisations and users already have Storage arrays (SAN,NAS,..) and migrating to Gluster is not possible or reliable. While I have 2 storage arrays I would like to use them like network raid 1. Also the combination of more storage types would be super cool (NFS + ISCSI) or so. Jonathan: Yes, I can rely on disk arrays for hdd fault tolerance, but there are more points of failure than disks (network, ram, cpu, water in rack and so). To be clear, I know, I can manage fault tolerance by hardware, but it is more expensive
On Thu, May 2, 2013 at 11:18 AM, Jakub Bittner wrote:
To be clear, I know, I can manage fault tolerance by hardware, but it is more expensive
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
If I understood correctly the initial thread message, Jakub's main concern is that if he creates a datacenter, then all storage domains have to be of the type initially chosen Taking from rhev 3.1 docs: " The type of data domain dictates the type of the data center and cannot be changed after creation without significant disruption. All storage in a data center must be of one type only. For example, if iSCSI is selected as the type, only iSCSI data domains can be attached to the data center. " So if he would like to put iSCSI and SAN LUNS on the same DC, he can't at the moment. If I remember correctly in another past thread it was said (Itamar perhaps?) that there wasn't a particular restriction or design constraint about this limitation, but probably only historical reason.... is this true? So it could be cool to put on the roadmap the chance to mix several SD types in the same DC. BTW: Jakub which is the kind of replication / network raid 1 you are using or planning, perhaps drbd or similar? Gianluca
Dne 2.5.2013 11:45, Gianluca Cecchi napsal(a):
On Thu, May 2, 2013 at 11:18 AM, Jakub Bittner wrote:
To be clear, I know, I can manage fault tolerance by hardware, but it is more expensive
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users If I understood correctly the initial thread message, Jakub's main concern is that if he creates a datacenter, then all storage domains have to be of the type initially chosen
Taking from rhev 3.1 docs: " The type of data domain dictates the type of the data center and cannot be changed after creation without significant disruption. All storage in a data center must be of one type only. For example, if iSCSI is selected as the type, only iSCSI data domains can be attached to the data center. "
So if he would like to put iSCSI and SAN LUNS on the same DC, he can't at the moment.
If I remember correctly in another past thread it was said (Itamar perhaps?) that there wasn't a particular restriction or design constraint about this limitation, but probably only historical reason.... is this true?
So it could be cool to put on the roadmap the chance to mix several SD types in the same DC.
BTW: Jakub which is the kind of replication / network raid 1 you are using or planning, perhaps drbd or similar?
Gianluca
To be honest, I did not know, that datacenter has to have only one type of storage domain. If so, you are right that it would be cool to mix several types. (NFS with ISCSI and so) But my first idea was to connect datacenter with two iSCSI storage domains and have something like raid 1 over them for fail tolerance storage ;-)
------=_Part_1213_14946045.1367495377477 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit If you use DRDB for that you must be sure that all virtual disks are only in one storage domain, and let DRDB mirror them to the other storage domain, right?. Can we mix 2 iSCSI storage domain with DRDB? for network based raid-1. Jose ----- Original Message ----- From: "Jakub Bittner" <j.bittner@nbu.cz> To: "Gianluca Cecchi" <gianluca.cecchi@gmail.com> Cc: users@ovirt.org Sent: Quinta-feira, 2 de Maio de 2013 11:49:33 Subject: Re: [Users] Fault tolerant storage Dne 2.5.2013 11:45, Gianluca Cecchi napsal(a):
On Thu, May 2, 2013 at 11:18 AM, Jakub Bittner wrote:
To be clear, I know, I can manage fault tolerance by hardware, but it is more expensive
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users If I understood correctly the initial thread message, Jakub's main concern is that if he creates a datacenter, then all storage domains have to be of the type initially chosen
Taking from rhev 3.1 docs: " The type of data domain dictates the type of the data center and cannot be changed after creation without significant disruption. All storage in a data center must be of one type only. For example, if iSCSI is selected as the type, only iSCSI data domains can be attached to the data center. "
So if he would like to put iSCSI and SAN LUNS on the same DC, he can't at the moment.
If I remember correctly in another past thread it was said (Itamar perhaps?) that there wasn't a particular restriction or design constraint about this limitation, but probably only historical reason.... is this true?
So it could be cool to put on the roadmap the chance to mix several SD types in the same DC.
BTW: Jakub which is the kind of replication / network raid 1 you are using or planning, perhaps drbd or similar?
Gianluca
To be honest, I did not know, that datacenter has to have only one type of storage domain. If so, you are right that it would be cool to mix several types. (NFS with ISCSI and so) But my first idea was to connect datacenter with two iSCSI storage domains and have something like raid 1 over them for fail tolerance storage ;-) _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users ------=_Part_1213_14946045.1367495377477 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: quoted-printable <html><head><style type=3D'text/css'>p { margin: 0; }</style></head><body><= div style=3D'font-family: arial,helvetica,sans-serif; font-size: 10pt; colo= r: #000000'>If you use DRDB for that you must be sure that all virtual disk= s are only in one storage domain, and let DRDB mirror them to the other sto= rage domain, right?. Can we mix 2 iSCSI storage domain with DRDB? for netwo= rk based raid-1.<br><br>Jose<br><br><hr id=3D"zwchr"><div style=3D"color: r= gb(0, 0, 0); font-weight: normal; font-style: normal; text-decoration: none= ; font-family: Helvetica,Arial,sans-serif; font-size: 12pt;"><b>From: </b>"= Jakub Bittner" <j.bittner@nbu.cz><br><b>To: </b>"Gianluca Cecchi" <= ;gianluca.cecchi@gmail.com><br><b>Cc: </b>users@ovirt.org<br><b>Sent: </= b>Quinta-feira, 2 de Maio de 2013 11:49:33<br><b>Subject: </b>Re: [Users] F= ault tolerant storage<br><br>Dne 2.5.2013 11:45, Gianluca Cecchi napsal(a):= <br>> On Thu, May 2, 2013 at 11:18 AM, Jakub Bittner wrote:<br>><br>&= gt;> To be clear, I know, I can manage fault tolerance by hardware, but = it is<br>>> more expensive<br>>><br>>> __________________= _____________________________<br>>> Users mailing list<br>>> Us= ers@ovirt.org<br>>> http://lists.ovirt.org/mailman/listinfo/users<br>= > If I understood correctly the initial thread message, Jakub's main<br>= > concern is that if he creates a datacenter, then all storage domains<b= r>> have to be of the type initially chosen<br>><br>> Taking from = rhev 3.1 docs:<br>> "<br>> The type of data domain dictates the type = of the<br>> data center and cannot be changed after creation<br>> wit= hout significant disruption. All storage in a data<br>> center must be o= f one type only. For example, if<br>> iSCSI is selected as the type, onl= y iSCSI data<br>> domains can be attached to the data center.<br>> "<= br>><br>> So if he would like to put iSCSI and SAN LUNS on the same D= C, he can't<br>> at the moment.<br>><br>> If I remember correctly = in another past thread it was said (Itamar<br>> perhaps?) that there was= n't a particular restriction or design<br>> constraint about this limita= tion, but probably only historical<br>> reason.... is this true?<br>>= <br>> So it could be cool to put on the roadmap the chance to mix severa= l SD<br>> types in the same DC.<br>><br>> BTW: Jakub which is the = kind of replication / network raid 1 you are<br>> using or planning, per= haps drbd or similar?<br>><br>> Gianluca<br><br>To be honest, I did n= ot know, that datacenter has to have only one type <br>of storage domain. I= f so, you are right that it would be cool to mix <br>several types. (NFS wi= th ISCSI and so)<br><br>But my first idea was to connect datacenter with tw= o iSCSI storage <br>domains and have something like raid 1 over them for fa= il tolerance <br>storage ;-)<br>___________________________________________= ____<br>Users mailing list<br>Users@ovirt.org<br>http://lists.ovirt.org/mai= lman/listinfo/users<br></div><br></div></body></html> ------=_Part_1213_14946045.1367495377477--
On 5/2/2013 5:18 AM, Jakub Bittner wrote:
Dne 1.5.2013 17:25, Sven Knohsalla napsal(a):
Hi Jakub,
I assume the solution you are searching for is something GlusterFS can serve for example: http://www.gluster.org/about/
Best, Sven.
-----Ursprüngliche Nachricht----- Von: users-bounces@ovirt.org [mailto:users-bounces@ovirt.org] Im Auftrag von Jonathan Horne Gesendet: Mittwoch, 1. Mai 2013 17:16 An: Jakub Bittner; users@ovirt.org Betreff: Re: [Users] Fault tolerant storage
I would be surprised if you could do that without your users noticing.
It seems like you would be better served to have each storage domain on a RAID array and rely on the hardware tools of the RAID to handle disk failure tolerance.
Cheers, jonathan
-----Original Message----- From: users-bounces@ovirt.org [mailto:users-bounces@ovirt.org] On Behalf Of Jakub Bittner Sent: Tuesday, April 30, 2013 9:19 AM To: users@ovirt.org Subject: [Users] Fault tolerant storage
Hi,
would it be possible in near future to use for example 2 attached DATA storage (iSCSi, or so) as fault tolerant storage?
I mean I have two data storage connected to data center and they include exactly same data and my VMs runs from one and all changes are mirrored to second like RAID 1 and if one storage fails everything switch to second storage without user notice? _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
________________________________ This is a PRIVATE message. If you are not the intended recipient, please delete without copying and kindly advise us by e-mail of the mistake in delivery. NOTE: Regardless of content, this e-mail shall not operate to bind SKOPOS to any order or other contract unless pursuant to explicit written agreement or government initiative expressly permitting the use of e-mail for such purpose. _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users Hi,
thank you for replies. Sven: I know Gluster. I am using it since 2009 in production but the problem is that many organisations and users already have Storage arrays (SAN,NAS,..) and migrating to Gluster is not possible or reliable.
It seems to me that the easiest way to get mirrored storage would be to use mirrored LVM logical volumes. That would require modifications to vdsm at least (probably the engine too for management) but certainly seems like it would be the easiest/most reliable way.
While I have 2 storage arrays I would like to use them like network raid 1. Also the combination of more storage types would be super cool (NFS + ISCSI) or so.
Jonathan: Yes, I can rely on disk arrays for hdd fault tolerance, but there are more points of failure than disks (network, ram, cpu, water in rack and so).
To be clear, I know, I can manage fault tolerance by hardware, but it is more expensive _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
participants (7)
-
Gianluca Cecchi -
Jakub Bittner -
Jeff Bailey -
Jonathan Horne -
suporte＠logicworks.pt -
Sven Knohsalla -
Vered Volansky