To Cluster, Replicate or do both for servers in US/Asia/Europe?

  • I know next to nothing about clustering and replicating, never had to do it. Currently we have a server in the US and are looking at putting one in Asia and Europe. I would like the data to be replicated on both servers maybe even cluster them to use as fail over for each other. Are there any issues with clustering the servers and replicating given the locations? Should I only replicate or should I do both if possible. Are there performance issues with clustering and replicating? If I cluster would I need to replicate?

    If I had the choice of SQL2000 or SQL2005 which would you recommend? 

    If you have links to clustering / replicating for dummies, best practices, how to or lessons learned it would be greatly appreciated!

     

  • It's hard to say without knowing more about your particular situation.  Is this for Disaster Recovery, Load Balancing, OLTP or OLAP systems, how many servers do you have available and the list goes on.  If you cluster, your machines should be located near each other, with replication you aren't limited in distance but bandwidth from Asia to Europe may become an issue.

    Have you looked up Books Online? 

    http://msdn.microsoft.com/library/default.asp?url=/library/en-us/startsql/getstart_4fht.asp 

    I always find that's the best place to start.

    As for SQL Server 2000 or 2005 - it depends how long you want to wait to implement this, given that 2005 won't be released until at least Nov.

  • Thank for the response.

    "If you cluster, your machines should be located near each other" - Clustering is out then unless we simply cluster here in the US to balance the increased load from Europe and Asia and what ever the users web application response times end up being that is what they get.

    "with replication you aren't limited in distance but bandwidth from Asia to Europe may become an issue" - Is there a rule for bandwidth to follow? Has anyone done this?

    RE books: I have grabbed a couple articles but have not had a chance to go through them all yet.

    I wonder what other companies that support global web applications do. What server configurations is setup? What types of response time are acceptable?

    My apologies for my lack of knowledge, some of my questions may not make much sense.

     

  • You need to be clear on what you're trying to achieve and then decide on the best way to achieve it.  I'm unsure if you're trying to perform load balancing or disaster recovery.

    At my site we have an Active/Active SQL Server & Analysis Services cluster on a SAN.  If Server1 goes down, all traffic is redirected and serviced by Server2 likewise if the reverse happens.  This is part of our disaster recovery architecture - it's not a load balancing arrangement.  We do have load balancing across our web servers for Reporting Services though.

    I don't have any experience in supporting global networks, so I can't really help you there.  Sounds like a big project you're about to undertake.  Do you have an Architect and Network Admins you can discuss this with?

    Angela

  • Basically I want to support our customers as best as possible. The servers would more so to ensure the response times globally are adequate as opposed to disaster recovery, if I could get both great. Disaster recovery could be handeled with backups and transaction logs.

    Our IT is supported by IBM but I thought if I could gain some knowledge and experience prior to talking with them it might help.

     

Viewing 5 posts - 1 through 4 (of 4 total)

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