July 5, 2006 at 9:20 am
I am wondering how many companies (if any) have identical hardware and software for their production and staging/testing environment.
For example where I work we have 4 primary production servers in two cluster groups. Each server has 4 processors.
Our staging environment has 2 primary servers that represent 2 of the 4 primary production servers. The servers in staging only have 2 processors and are not clustered. Plus they are older machines than the ones in production and not the same brand.
So how about where you work? Is it unrealistic for me to request that our staging envirnment be upgraded and expanded to more closely match our production environment?
Robert W. Marda
Billing and OSS Specialist - SQL Programmer
MCL Systems
July 5, 2006 at 12:03 pm
it really depends on what type of testing you plan on doing.
Our test environment (dual 2.4 w/ 4 gig) is not as beefy as our production environment (quad 3.2 w/8gig). We have developed baseline tests against our test environments and then compare these performance results with what we get from subsequent software releases. THey are different than what we would get from production but give us an idea of how the changes will perform.
If you do plan on doing some full scale capacity/load testing you would need to have very similar hardware. I"m not sure how many folks could afford such a test environment. We run our capacity tests against our DR environment (which happens to be the same as prod).
I hope this helps.
July 5, 2006 at 1:15 pm
At one place I worked the difference between the testing and production hardware was that the testing hardware was slightly more beefy than the production environment. The idea was that testing would be a damn site more intensive than the real thing.
Otherwise the servers were the same model, same OS, same patch level.
With SQL2000 I would be nervous about having test/production servers with too big a difference in RAM. For example SQL will deal with 2GB automatically, but 8GB needs the PAE switch and other configuration options.
Similarly I wouldn't worry about a Dual vs Quad processor machine but I believe that moving beyond Quad processors does introduce configuration issues into the mix.
I think "The best of SQL Server Central 2003" had an article on maximum degree of parallelism in queries that showed that beyond a certain number of processors queries could end up running slower.
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply