MS SQL Server Performance/Redundancy?
Ok, a little background. We have about 1000 clients that hit our servers hard. We have 7 or so webservers and 4 database servers. Right now, if a database server has to be rebooted, all clients on that machine are down for the length of the reboot. Even scarier, if anything on any of the machines dies, all the clients on that database server is down until we fix the server! Other times, a single server for a group of clients gets overloaded and we currently move their db to another server. We need something that can let the extra load spill over to another, less busy machine.
We need redundancy somehow. I want to be able to upgrade hardware on the machine, install MS updates, or know that if a cpu dies that our clients won't be left in the dark. What can help this? Clusters? Replication?
Will replication work? If so, how would you tell the scripts on the webservers to look at another database server if 1 went down?
I know clusters can be super expensive if you get a SAN, but is there another way that will work with less money? A machine with raid 0+1 for 500gb hd redundant space that does nothing but filesharing with gigabit ethernet to each machine?
I'm not limiting the possible solution to clusters or replication. If you have any other suggestions, please, throw them out.
Thanks.
|