Designing a new system

  • I have been asked to design a new high availability SQL server system and I was looking for some opinions. I currently only have one production server and one backup server, both running Windows 2000 server and SQL Server 2000 Standard Edition with transactional replication running between them. I have a main database currently at 30 GB which will likely grow to approx 150 GB in the next few years. We are not a 24X7 shop. Most of our OLTP processing is done at night and mainly reporting during the day.

    I am thinking I will get a couple of servers running Windows 2003 server and SQL Server 2005 Standard Edition with some sort of disk array for the main data storage. I was trying to decide between using Clustering or a Standby Server. Any opinions would be greatly apreciated. Thanks for your time.

  • If you're planning on using Standard edition, aren't you limited to Standby server? I thought clustering was only available in the Enterprise Edition?

    I think in the end it comes down to how much down time you can afford.

    I believe with clustering the down time is zero (unless every server in your cluster dies) as all servers are constantly in place. A good option if you need to scale out too I bleieve.

    With Standby you will loose time while the Standby server comes online - not sure how much though.


    Regards,

    Steve

    Life without beer is no life at all

    All beer is good, some beers are just better than others

  • With SQL 2000 you needed to buy enterprise if you wanted to do clustering, however, with SQL 2005 you can now do a 2 node cluster with the STD version.

    If you need more nodes then that the ENT edition supports up to 8.

    If cash isn't too tight then i'd go for clustering, and if you need off-site recovery too then i'd go for clustering with remote standby via log shipping.

  • I would also say go with clustering and logshipping combo. Clustering is great for a hardware failure but if the database goes belly-up and you need quick recovery the logshipping will handle that. Also with clustering using a SAN there is a single point of failure if the SAN goes down (which is probably a very remote occurence but you need to think abaout it).

  • Yep, we use a SAN for this, whilst i agree it is a single point of failure there's generally a good amount of redundancy built in (failover disks, multiple paths, etc).

     

  • Thanks for all of the suggestions. I believe I will use the 2005 STD edition and do a 2 node cluster with log shipping to a remote site for disaster recovery purposes.

    Does anybody use any backup/compression software other than native SQL backups? Any advice/suggestions on this topic?

    Thanks for your responses

  • LiteSpeed 2005 and SQL Backup are both really good at this sort of thing, I have no problems with recomending either of them to you.

     

Viewing 7 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic. Login to reply