April 29, 2010 at 6:38 am
Background :
we have like 50 huge databases residing on 2 production servers but we would like to bring all 50 into one server but before that i would like to do query performance against 2 servers Vs 1 server.
All our production databases are used by analysts who has only read access to these servers and there would be no other users or processess running on them.most of the time they run queries which hit CPU usage to 100% as they run queries on multiple databases at the same time.
I have collected few queries from the users and planning to test them by embedding their queries into procedures and shcedule them over the week end, by that way i can check duration of queries executed on 2 servers Vs 1 server.
Is there any other best way to test which helps me to decide whether or not to go with the plan ?
April 29, 2010 at 8:40 am
Tara-1044200 (4/29/2010)
Is there any other best way to test which helps me to decide whether or not to go with the plan ?
Run PerfMon and a server-side Profiler trace on your current production system. Get baseline values for ALL hardware and software that's currently running.
Then take the Profiler Traces and replay them against your single server while running a new server-side trace and PerfMon trace on that server. This should give you a baseline of how traffic will act differently on the single server compared to the 2 server method.
April 29, 2010 at 6:17 pm
Something to consider when you set up test environment - make sure you use the same backup to setup databases on servers you are going to compare. You can have completely different performance results on test server comparing to another test server (identical servers) when indexes have different fragmentation level.
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply