March 25, 2011 at 7:16 am
Hello guys,
I have a question. I think it's possible but I want to be sure before moving on with this solution.
Maybe you can help me.
I want to know if it's possible to make the next configuration:
3 physical servers with OS and SQL clusters configured in the following way:
1 - with one active node X
1 - with one active node Y
1 - with both passive failover nodes XX (for node X) and YY (for node Y)
Does anyone have this configuration or can you tell me if it is possible without involving big issues?
Thank you very much
Wish you good ideas! 🙂
Andreea
March 25, 2011 at 8:50 am
Yes this is possible, and i would imagine it is quite common when trying to get the most out of hardware without having too many boxes sitting unused.
Essentially ehat you have here is two a/p clusters.
Lets say nodes 1-3.
Setup your two Windows clusters to use the same node (3) as a cluster member.
That way, when clusters failover they will fail to node 3!
Be aware that in t he scenario that both clusters (sql instances) failover at the same time, all you processing is going to happen on the same server so there will be a performance drop! You might want to keep this in mind when buying memory for the passive or setting max mem for the actives.
Its a bit obvious but you cant have D: and E: drives on both actives if they could potentially share the same physical space. So keep that in mind also.
Good luck and let me know how is goes (as i said it is not only possible but its quite common, so there should be tons of threads out there).
Adam Zacks-------------------------------------------Be Nice, Or Leave
March 25, 2011 at 9:07 am
Hi,
Thank you very much for your answer. I've found some posts with 3 nodes configuration (2 passive and 1 active) but there were different scenarios and that's why I've asked here...
Yes, I know it's quite unpleasant if same clusters fail in the same time and that's why the machine that holds the passive node should be stronger...
I don't know when we'll implement this (not quite soon), but when we'll do it, I'll tell you how things worked and what issues we had.
I have to search more in order to see all the details about this configuration.
Thanks
Wish you good ideas! 🙂
Andreea
March 25, 2011 at 10:02 am
To make it easier on youself, think of it as two a/p's.
Then when it comes to configuration add the same 3rd node to each.
Just make sure to think about machine names and instance names. So maybe 'SQLCLUSNODE*' for the nodes and 'SQLCLUSTER1' and 'SQLCLUSTER2' for the SQL virtual name and maybe for the SQL instance name 'SQLINST1' and 'SQLINST2'.
Keep in mind that each isnstance has to be a named instace so and the string must be made up of the SQL virtual name and instance name; so in this example your connection string would be 'SQLCLUSTER1\SQLINST1' etc.
Dont make it too complex else you users will get confused. We have the same instance name as virtual name so SQLCLUSTER1\SQLCLUSTER1 is much easier to remember.
Adam Zacks-------------------------------------------Be Nice, Or Leave
March 25, 2011 at 11:53 am
We have an active / active / active / passive setup. So, yes, definitely possible. Use virtual names when possible so when your servers reach EOL, you're not changing a zillion config files, SSIS folders, and jobs when you move things over.
DB1, DB2, DB3, etc.
The major things to watch out for is to
A) make sure all your services that are cluster dependent are set up in Cluster Administrator in the proper group,
B) Make sure all your shares are set up in the Cluster Administrator (so they properly fail over),
and C) Make sure you use Share Security, not drive / folder security so that when the SAN flips, you don't have to redo your security every time the cluster fails.
Oh, and D) Use different names for your SSIS folder (if storing packages in MSDB instead of the file system) so when both nodes fail over to the passive, your jobs don't break. If you name that folder "PROD" for both nodes 1 & 2, the day both of them fail over to node 3 at the same time, those folder names (because you can't use the same name twice) will rename themselves to Instance1 and Instance2, breaking any SSIS jobs that use the path "\PROD\Packagename.dtsx". Tis a PAIN.
March 29, 2011 at 2:56 am
Hello again,
So... before buying 3 servers in order to do the configuration above (1st server - active node for one customer, 2nd server - activer node for the other customer and the 3rd one with 2 passive nodes for both customers), can I test this configuration on 3 virtual servers?
Even if we have now a cluster configuration, I don't know how to do it (I haven't done this before). Where can I find how to set the clusters and how to set two clusters on the same machine and how to share the common resources?
Thank you
Wish you good ideas! 🙂
Andreea
March 29, 2011 at 3:49 am
As far as the Windows clusters are concerned, I dont know (we have a seperate team that does that).
As far as SQL cluster is concerned this should give you a good start. It is based on 2005 but there shouldnt be much diff.
Yes I believer you can test this i n virtual if you have the resources. If its your first time probably best. Watch the MSDTC config (cluster) it can be tricky.
HTH
Adam Zacks-------------------------------------------Be Nice, Or Leave
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply