SP2010 minimum hardware req on vSphere

Question

I had a read on https://www.vmware.com/files/pdf/solutions/SharePoint_2010-Best-Practices-Guide.pdf and there are some statements like (page 10)

"Allocate the minimum requirement for production virtual machines based on Microsoft guidelines, the role of the virtual machine, and the size of the environment."

then

"SharePoint 2010 minimum processor requirements recommended by Microsoft may be excessive in some environments. For this reason, VMware recommends reducing the number of virtual CPUs if monitoring of the actual workload shows that the virtual machine is not benefitting from the increased virtual CPUs."

Then again on page 38 "Note the new SharePoint 2010 minimum requirements, for example, all server roles are now required to have at least four 64-bit processor cores (or vCPUs)."

So when they say that Microsoft guidelines could be excessive does that imply for anything above the minimum requirements (four cores) or do they mean that go down to 2-3 cores would be OK if the response time is OK?

Thanks in advance.

Answer 1

As it reads, if you see an underutilization of CPU you can remove one virtual core and see if your environment is still handling the load.

This is nothing specific to vSphere or even SharePoint, it is how you normally find optimal performance from just enough resources (by adding or removing resources and measure the outcome)

Answer 2

SharePoint performance depends heavy on the functionality and how you implement your solution and topology. Certain functionality may demand a lot more resources. I have seen 1 server installations go fast and a 5 server farm go slow.

In general for a production envirnoment I would go for the 4 cores. But my personal development machine for example runs smoothly on my laptop with only one core.

Also keep in mind that SharePoint has quite a lot of background jobs schedueled. They might influence the performance as well. The crawl (indexing) job of the search is a typical long running resource intensive process that runs at a certain time or intervals. Users may experience slower response times during the executing of that job. (this also depends on the topology of your setup)

It's all about the expected load and what to do during peak times and/or heavy utilisation.