August 3, 2004 at 7:15 pm
I was reading and article on clustering and read that clustering does not offer data fault tolerance, what is meant by that exactly?
Also if I were to run a two node cluster in an active / active configuration would I need two SQL Server Enterprise licences one for each server?
If I have a dual processor SQL Server, I know I need to licence both processors but do I need to licence four processors in a two node cluster?
August 3, 2004 at 11:42 pm
Data Tolerance means , if one instance fails, e.g. the active server gets dis-connected from the network or crashes, then the other[passive node] will become active and take the load. This should happen automatically and users may get disconnected for a minute or less, but when connected again they'll be served by the other server. This is called Failover Clustering.
If you run Active/Active then you need a license for each active instance. So , if you have Active/Active cluster on 2 servers with 2 CPU each, then you pay for 4 CPUs ...
August 4, 2004 at 9:46 pm
Data fault tolerance... in other words, the operating system itself does nothing specific to protect the data just because it's a cluster. You usually build fault tolerance in with the use of hardware raid arrays. Still, the hardware raid then represents a single point of failure (though the likelihood is rather minute) and if it goes, you've lost your data.
Contrast this with a system that might store the exact same data across several distinct arrays with the capability to switch to another array should one fail. That would be data fault tolerance. SQL Server clustering has no support for such a concept.
K. Brian Kelley
@kbriankelley
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply