September 12, 2012 at 1:47 pm
Hi guys,
I have a sql 2008 r2 enterprise box which is running on windows 2008 R2 enterprise and the business wants to separate some of the databases from this instances and the recommendation that has been given to me is to have another sql instance installed in the Active\Passive cluster and move the databases over.
What do you guys think are the pros and cons on this set-up?
Eventhough the new instance wwill have its own cluster group wouldn't it share the processor and ram? Wouldn't an issue with one of the partitions cause both the instances to fail over?
Your input will be greatly appreciated.
September 12, 2012 at 2:43 pm
wannabe dba (9/12/2012)
Hi guys,I have a sql 2008 r2 enterprise box which is running on windows 2008 R2 enterprise and the business wants to separate some of the databases from this instances and the recommendation that has been given to me is to have another sql instance installed in the Active\Passive cluster and move the databases over.
What do you guys think are the pros and cons on this set-up?
You typically move to multi instances when you have a requirement for
wannabe dba (9/12/2012)
Eventhough the new instance wwill have its own cluster group wouldn't it share the processor and ram?
Yes it would but you can restrict these resources at each instance level
wannabe dba (9/12/2012)
Wouldn't an issue with one of the partitions cause both the instances to fail over?Your input will be greatly appreciated.
No, the shared disks are unique to each instance. A disk issue for instance A would have no effect on instance B.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
September 13, 2012 at 8:57 am
If different applications in a single instance (or sub-applications within a single application) are stepping on each other, then setting up a second instance can be an effective way of dealing with the situatuation. You could limit CPU with processor affinity or WSRM, and memory can be allocated based on need and prority. While resource governor could also be used, it is sometimes hard to track down ALL of the processes that could be problematic, and it only takes one that got away to make your life difficult.
Availability is a two edged sword. If there is a problem that causes one instance to be recycled frequently, it's nice to have part of the user population not affected. OTOH, if a cluster reboot is required, i.e. bacause Microsoft says so after you have opened a PSS incident, the other users might be upset if the problem hadn't impacted them. Having supported shared clusters for about a decade, I have heard 'why do we have to suffer an outage because of them?' more times than I care to remember.
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply