Hello,
I'm currently configuring a SQL server in Azure as a IaaS and had a questions about available IOPS to the database volumes. I have two options in front of me that I have laid out in detail below.
I am building storage volumes from Azure disks which are inherently resilient so no RAID5/6/10 etc is needed. There are 16 disks on the server. Each disk has an IOPS cap of 500. the IOPS scale linearly as you add more striped azure disks.
I have 4 volumes I will need in the SQL server. They will be F,G,H,I (Data,Logs,Temp,Install/Programs/Backups respectively). The server will host Dynamics CRM and DynamicsGP for about 100-150 users.
Method One:
RAID 0 across 16 disks resulting in 8000 IOPS on the single 16TB volume. (16x500=8000)
Slice that volume into the 4 needed drives.
That results in 8000 IOPS max but also shared across all volumes.
...