SQL Server Performance Forum – Threads Archive

Clustering 2005, is it for me?

I’ve been assigned to a project where I need to have high availability for a Asp.Net with MSSQL 2000 or 2005. We have two Dell servers that are of medium quality. Unfortunately that’s where the budget stops. We do happen to have iscsi laying around so I figured I’d throw together a test environment (while watching excepts from kb89179<img src=’/community/emoticons/emotion-11.gif’ alt=’8)’ /> and followed the process for installing the SQL server. I had to make the nodes AD’s (which was done prior to the install of CS and SQL2005). I’ve got a about 5 hours into it and I’m wondering if this is the path that I need to go down. We need high availability DB only. The web app is stateless and can actually be on both nodes at the same time. Unfortunately I really can’t afford to spring for a iscsi array for this project (mostly out of pocket myself for an NPO). So I have two questions… 1) If I want to continue to use the cluster similar to that above, is there a way to do it without the shared array. I have read various acticles that talk about local mirrored qorum files. This would be optimal. 2) Do I even need to cluster this? I was thinking of Merge replication and have each node just use localhost and sync internally and use Windows Load balancing to do the rest. The machines are Dual Xeon Dell’s with 4gb ram each and mirrored 250gb SATA drives. Any advice would be greatly appreciated.

Clustering is complex and requires specific, certified hardware. If you don’t have it, don’t try clustering. If you are not in a huge hurry, consider SQL Server 2005 disk mirroring. While 2005 had disk mirroring in it now, it won’t be fully supported until the first service pack comes out for 2005 in the middle of this year. It provides all the needs you have described and does not require any special hardware. Check out Microsoft’s website on disk mirroring to learn more. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com

brad is almost correct, but he’s talking about database mirroring and also SP1 is in CTP at the moment. Disk mirroring is a completely different kettle of fish. Database mirroring protects that database and work in several modes – without a third box to do witnessing though you won’t have automatic failover and can only run in High Protection mode.
The next question is really what data is held and how does the web server and users interact with it?

]]>