I'd like to get some perspectives on performance questions related to sizing your environment and to capacity planning.
Here are the issues:
I want to look at the performance demands of my enviornment, what VMs are demanding now, and predict what they are going to demand as I add VMs.
As far as what I have now, I can see what my demands are. I can look at CPU demand, memory demand, IOPs demand, network throughput demand for all my VMs. I can see what an "average" VM uses in each of these categories. I can also see what an "average" SQL server, web server, application server, or file server uses and create a profile that estimates what a server of this type will demand in the future.
Some challenges are:
Preparing to Implement self provisioning -
It becomes harder to know who is going to request what kind of VM and when, and how many Vms are going to be requested.
I want to get as much and as accurate "baslining" and performance data as I can to prepare for self-provisioning.
Is there a standard practices or good articles that would discuss how to thoroughly do the sizing and analysis of my current VMs to best prepare for the future unknowns?
(vCOps is available). Should I create a profile for each application type? Should I simply estimate how many new VMs will be created, and use my historical data to estimate the increase in performance demands?
