SQL Server Performance Forum – Threads Archive
Do I need clustering or something else?
I have a SQL server with around 200 databases. I want to implement an enviroment with the following criteria: 1. Another SQL server with all the databases2. Failover so that if the primary SQL server fails the other will kick in
3. If I create a database on one server it will automatically be created on the other
4. If I change for example the type of recovery mode on one database on one server it will be changed on the other Does anyone have suggestions on the the technology used for this? Thanks a lot, Dennis
With clustering you can achieve failover capabilities and of course High Availability.
If you need to replicate on another server all your actions, you may need to have a look at Double-Take at:http://www.nsisoftware.com/
HTH Franco
Define your requirements, not the solution. Why do you want another server?
What is the SLA you are trying to achieve?
Why do you want actions duplicated?
Why do you want another server? In case the initial server fails. What is the SLA you are trying to achieve? 99.997 uptime Why do you want actions duplicated? So I do not have to do them on the other server.
Another possibility is log shipping to another server, wiht load balancer in front of both db servers.
I think you missed my point, that you are deciding half of what you need and then asking for a solution. Given the only useful information you’ve posted, the SLA, you’re looking at roughly 4-5 hours downtime per year. I would recommend a four node cluster to achieve this. Your modifications to SQL Server will only need to be made once as there is only one instance of SQL Server, but your modifications to the OS wil usually have to be replicated on each node. In this configuration you are allowing for hardware and OS failure.
From our 200 database we have 10 of using SQL replication. Our SQL server is the Publisher and Distributor and the client plays the Subscriber role. Will you clustering solution work with replication? Thanks, Dennis
I haven’t tried it, it might be complicated but I can’t see a reason that it wouldn’t work, here’s some info:http://www.sql-server-performance.com/q&a24.asp Just check that you aren’t invalidating your reasons for replication when you move to a cluster and try to keep your replication models as simple as poss.
]]>