|
Hi we have 2 Windows 2008 standalone servers that we want to cluster with nlb and then install SQL Server 2005 on. Will or can this work with SQL Server 2005? We don't want to create a domain to create a failover cluster. I know this is a weird setup, but that is how it needs to be configured. Thanks in advance
(comments are locked)
|
|
Microsoft claim that this will work, and promote peer-to-peer replication as the mechanism. The peer-to-peer article in the SQL Server 2005 Books Online -- http://msdn.microsoft.com/en-us/library/ms151196(SQL.90).aspx -- shows the layout for using an NLB cluster in front of two, or more, SQL Server 2005 servers that use peer-to-peer replication to stay synchronized. Certainly - +1. The most important part, though, is to remember that you need to set up replication between the load balanced servers - there isn't a 'magic NLB' feature.
Nov 06 '09 at 07:48 AM
Matt Whitfield ♦♦
I don't see how this can actually be touted as load balancing. The 'write' load is applied equally to all the participating servers, so in essence each server needs to be able to cope with the load in isolation.
Nov 13 '09 at 04:34 PM
Kev Riley ♦♦
That's a good observation. This solution only balances the read load. Each server still has to be able to cope with the 'write' load in isolation.
Nov 13 '09 at 05:55 PM
KenJ
(comments are locked)
|
|
What are you trying to achieve? A load-balanced SQL Server? It won't work like that. High-availability solution - see this question for an alternative. my it mananger seems to think that you can load balance 2 sql servers via the nlb feature. We're busy setting it up, but I can't see how that will work. The main thing is that we try to avoid creating a domain, in which case you can setup a failover cluster on windows 2008
Nov 05 '09 at 02:01 PM
Danie
think of it this way: if you load balance a database, and one transaction writes some data to the 1st server, then a 2nd transaction issues a read to the 2nd server, how does the data get from one server to another? You need to either share-disk (generally a cluster with a SAN, but still only one db engine), or share-nothing (multiple db engines and data silos , a standalone SOA type architecture). What is the ultimate goal? High-availability or scalability?
Nov 05 '09 at 02:21 PM
Kev Riley ♦♦
(comments are locked)
|
