[grepfru_it]

I wonder why they are not talking about the elephant in the room, the AWS spend. Cut back on that and these layoffs probably aren't necessary. The problem is the sheer size of GitHub data and the unreliability of Azure. There is an entire datacenter that is unused because data locality severely limits performance.

[umeshunni]

Moving from AWS to Azure is an order of orders of magnitude harder than moving from Zoom to Teams.

[grepfru_it]

My complaint is that MSFT had 3 years to focus on this problem and the solution today is to let go of engineers rather than prioritize minimizing costs (i guess in a way they did). There were EC2 snapshots dating back to 2013 when I last checked. Bad management gonna be bad

[bato]

Pretty much all new features from acquisition news are on azure (packages, codespaces etc).

Which is why outages usually match Azure outages.

[grepfru_it]

Hello fellow hubber? We still deploy new services using AWS or our own internal datacenters. Projects, for instance, is still run on k8s, not AKS. AE development was complicated from the start due to azure capacity, so much so that a tiger team went back to building it from scratch without azure.

Our AWS spending is outrageous tbh

[mardifoufs]

Are they using both AWS and Azure?

[grepfru_it]

Yes