There are several information which are required to make a reasonable sizing estimate for Azure VMs, and here is a comprehensive list which we used to follow from long time to get the reasonable T-Shirt sizing of the Azure VM’s.
- CPU utilization
- Memory Utilization
- Disk Size
- Disk IOPs (I/O’s per second) for C: drive and data / database drives
- Disk data transfer rates (Mbytes per second) for C: drive and data / database drives
- Backup / restore requirements
- Availability / redundancy approach
- DR approach
- Any specific network requirements (For example DMZ, Load balancer)
- VM scheduling opportunities – I.E. when can it be powered off to save money.
Out of these many parameters the first five items are the most critical for sizing as over-provisioning in Azure can have major cost implications. The first five can be easily gathered by configuring Windows Performance Counter. The information of the next 5 can be gathered by the discussion with application owners. In my future posts I will show you how we can capture the information of the first five parameters from Windows Server OS based VM’s or Physical Servers.