msql cluster server replacement

I want to upgrade my hardware in my sql cluster - is it possible to use imageing software and restore the image on the new server hardware - has anyone had any luck with this?
zingabAsked:
Who is Participating?
 
Aaron ShiloChief Database ArchitectCommented:
hi

i never had any luck with this since the hardware changes the image will never boot.

but i guess the querstion is how far are you goining with the upgrade.
0
 
zingabAuthor Commented:
what would be the best course of action?
0
Upgrade your Question Security!

Your question, your audience. Choose who sees your identity—and your question—with question security.

 
Marten RuneSQL Expert/Infrastructure ArchitectCommented:
Move over all databases to one node. Evict the other, set up new hardware for the first new clusternode. Add clusternode, install sql cluster up to same SP on this new clusternode. Failover/move all databases to the new clusternode. Evict the second OLD cluster node. Repet the steps to install hw, add clusternode, SQL, SP etc.

DONE!

//Marten
0
 
Ryan McCauleyData and Analytics ManagerCommented:
If your new cluster will have the exact same OS as well, you can do as marten suggests - essentially add your new hardware to your existing cluster, get SQL installed to exactly the same level, and then fail the databases onto the new hardware. Once that's done, evict your old nodes and you're done.

If you're upgrading the OS (we recently went from Windows 2003 R2 -> 2008 R2), there's no "upgrade path" per se - you have to actually move the databases. You can make this as smooth as possible, though, but moving all the logins (with SIDs intact) to your new database instance before you move your data over, using this script:

http://support.microsoft.com/kb/918992

This copies your logins, with passwords intact, to the new SQL instance. Once that's done, restoring your databases on the new instance will result in properly mapped logins and everybody can resume using the application immediately.
0
 
zingabAuthor Commented:
is there anything that you guys are aware of documenting this process step by step...we are keeping the os the same version - sql 2005 standard
0
 
Marten RuneSQL Expert/Infrastructure ArchitectCommented:
Standard is what I presumed, thats why you need to evict one node before adding the first hw, and agin evict the other old node before adding the second new hw. Standard ed only supports two node cluster.

Prerequsite is of course windows server enterprise edition, it has the ms cluster service.

Quote my posting above: "SQL, SP etc". This means documenting SP and hotfix level. New nodes needs to be in the same SQL SP/Hotfix level.

//Marten
0
 
Marten RuneSQL Expert/Infrastructure ArchitectCommented:
The command:

select
serverproperty('Edition') as SQLEdition,
Serverproperty('ProductLevel') as ServicePack,
ServerProperty('ProductVersion')as Version,
ServerProperty('MachineName') as ServerName

Should give you the information you need. The column version can be compared to: http://www.sqlservercentral.com/articles/Administration/2960/
0
 
Marten RuneSQL Expert/Infrastructure ArchitectCommented:
Oh, and the windows enterprise edition should also be on the same OS SP level as existing nodes. Felt I wanted to point this out to!

//Marten
0
 
zingabAuthor Commented:
thanks -  last question is there a site that shows the process that you guys know of?
0
 
zingabAuthor Commented:
Thanks everyone - now just getting back to my original though- would taking an image that is able to do a universal restore work?
0
 
Marten RuneSQL Expert/Infrastructure ArchitectCommented:
No images is bound to hw, chipset on the motherboard, nic drivers etc etc.

Thats why normally you put some hypervisor between the hw and the os, THEN you can easily change the hw later on. But thats another story.

No you cant successfully change hw using a snapshot. It might work but you wont gain optimal performance from your hw.

Regards Marten
0
 
Ryan McCauleyData and Analytics ManagerCommented:
Though there's (in theory) no reason that wouldn't work, since there is software designed to do that kind of swapout, I'd be very skeptical about it since Cluster communication between nodes is so sensitive to any kind of disturbance and the cluster is so tightly bound to the hardware. There's no reason you couldn't give it a shot, just to test it, on an isolated network and see if you can clone the disks, but I wouldn't have high hopes.
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.