In October of 2022, Microsoft Azure released Cosmos Postgresql, a distributed relational database. This competes directly with AWS (Amazon) serverless Postgresql which also allows an autoscaling solution that can span multiple nodes.
I'll get right to the offering. The lowest possible tier is 1 node with 2 vcore and 8gb of RAM and 128gb of storage. This ends up costing ~175 USD/month (350 USD/month with fail-over). Ouch. I am really surprised there is no 1 vcore or even 1/2 vcore option with lower RAM like AWS serverless provides. Hopefully Microsoft will fix this.
At the other end of the spectrum for the huge companies is the maximum of 20 nodes and 96 vcore per node and 672gb RAM per node along with 2tb storage per node. What's really baffling here is the low amount of maximum storage per node. To have 672gb of RAM per node but only 2tb of storage per node (only ~3x the RAM) is perplexing. I've emailed Microsoft but have not gotten a response. The storage maximum should be unlimited and charged per tb, or at least have an option for that. Very odd, but it is a new product so my guess is Microsoft will improve this over time.
The maximum configuration ends up being a whopping 213K USD/month. Double that (426K USD) if you want fail-over.
When more than one node is chosen, you also configure a coordinator node which adds to the price and would be painful for folks on a budget. For the big companies, this is likely a non-issue.
Microsoft touts the following features:
- automatic high availability
- built-in pgBouncer
- easy monitoring
- private endpoints
Until Microsoft addresses the storage limitation issue and the high price for someone just getting started with a minimal system, I would suggest folks stay clear of this offering for now. Microsoft is bound to clean some of this up over time and if they fix the pricing issues on the low end and storage on the high end, this will be a very competitive and compelling offering.
Source. Pricing determined by Azure Portal as of January 20, 2023.
Technomancer is a science and tech enthusiast who enjoys writing about software and AI and other tech topics.