On Sun, Mar 6, 2016 at 1:24 PM, Sunny Shin <sunny4s.git@gmail.com> wrote:

The relationship between ManageIQ and CloudForms seems the same as the one between oVirt and RHEVM, right?

Indeed. And we (the oVirt team) intend to contribute more and more to the ManageIQ oVirt provider, to make it best in class!
Here's a fresh example being worked on - https://github.com/ManageIQ/ovirt/pull/47
Y.


Thank you so much for all your hard work and good information. 

Sunny


2016년 3월 6일 일요일, Yaniv Kaul<ykaul@redhat.com>님이 작성한 메시지:

On Sat, Mar 5, 2016 at 4:48 AM, Sunny Shin <sunny4s.git@gmail.com> wrote:

It seems one engine has limitation such as not supporting over thousands or ten thousands of hosts. I heard that openstack supports unlimited(?) hosts with engine clustering similar to active-active ha. So, I thought that if active-active ha is supported, much more hosts are supported with one engine set. 

As for OpenStack, I'll just advise that you don't believe everything you hear. 


With current ovirt engine, if we want to deploy thousands of hosts, we may need multiple separate engines with hundreds of hosts each. And each engine has its own admin portal without one combined user interface. Am I right?

Correct, though you can and should use ManageIQ[1] for example to manage multiple oVirt engines. We are eagerly working on improving our integration with it.
We are looking both into the next generation of oVirt engine to allow higher scale as well as the coming release to improve the scale somewhat.
Y.


Sunny
 

2016년 3월 5일 토요일, Yaniv Kaul<ykaul@redhat.com>님이 작성한 메시지:

On Fri, Mar 4, 2016 at 11:48 AM, Sunny Shin <sunny4s.git@gmail.com> wrote:

Yaniv,

Could you a little bit more elaborate what you are working on for engine scailability? I thought that active-active ha is the only way for that.

Active Active is useful for both HA (where we believe hosted-engine is a good solution) as well as scalability. If we improve the current scalability, then we are good there too.
I've given several items below - it'd be difficult to enumerate all of them - I suggest going through the gerrit commits in the last 2 months or so and look at some of the work that has taken and is taking place, across all areas of the product. Few random examples:


Is there anything specifically you are looking for?


Sunny


2016년 3월 4일 금요일, Yaniv Kaul<ykaul@redhat.com>님이 작성한 메시지:



On Fri, Mar 4, 2016 at 10:45 AM, Sunny Shin <sunny4s.git@gmail.com> wrote:
Hi Yaniv,

Is there a plan to implement active-active HA in the near future? Is there any progress on it since the discussion in August 2013?

No, not at the moment. We feel that Self-Hosted Engine (SHE) is a great solution thus far.
We are working on improving the performance and scale of the engine, to be able to handle more load, more hosts, more VMs, etc.
This is an ongoing work across all components, from VDSM, to engine core, REST API, database calls and more.
Y.
 

Sunny

2016-03-04 17:38 GMT+09:00 Yaniv Kaul <ykaul@redhat.com>:


On Fri, Mar 4, 2016 at 2:03 AM, Sunny Shin <sunny4s.git@gmail.com> wrote:
Hi All,

In the new architecture page (http://www.ovirt.org/documentation/architecture/architecture/), overall architecture picture shows that engine supports active-active high availability.

This is inaccurate - the engine is not highly available in A/A architecture. I'd appreciate if you could file an issue for it so we'll fix this page.
 

However, as per engine HA page (http://www.ovirt.org/develop/release-management/features/engine/engine-high-availability/), active-active HA is not supported yet and there are several implementation issues.

So, could anyone give me a brief explanation about discrepancy between these two pages, and about the plan to implement this feature if not implemented yet?

I suggest looking at self-hosted engine, which provides non-A/A high availability for the engine.
Y.
 

Sunny

_______________________________________________
Devel mailing list
Devel@ovirt.org
http://lists.ovirt.org/mailman/listinfo/devel