New Reference Architecture: SQL Server on Azure VM's with Pure Cloud Block Store
This is a brand new, weeks-old reference architecture — and I’m really excited about this one.
During development, one of the most surprising discoveries was just how much Azure VM performance is limited by the IOPS cap tied to managed disks. It caught me off guard how much planning it takes just to size storage and compute together when you go the native route.
With CBS, I was able to bypass those constraints. It felt more like working with enterprise storage (which is what its meant to do!) , I could pull from a pool, scale performance independently of VM size, and provision storage volumes in a clean and easy way.
This new RA covers:
- SQL Server architecture on Azure VMs with Pure Cloud Block Store
- Snapshot-based backup and restore
- DR patterns using ActiveDR™ and HA using ActiveCluster™
- Dev/test database cloning with volume snapshots
- Performance benchmarking vs. Azure Premium SSD v2
It prooved:
- ~40% more transactional throughput (TPROC-C)
- ~93% better analytical query performance (TPROC-H) (using queries per minute normalization)
- 3–5x data reduction vs. raw data
Download the full reference architecture here
Would love to hear your thoughts on this architecture and how we could improve the expirience!
