Platform Teams Are Dead

Most big places I've worked, there's an 80% use case where devs want to ship a jar file and have a db available.

A central team usually provides a not-great but "ok" PaaS. All the external stuff (networking, sometimes certs, pipelines) is usually weird and slow, but it works eventually if you file enough tickets, and you get to hand off most support.

Then there's 20% of teams doing something special or big, who have political clout to bypass that. They ditch the broken org deps, and operate a "you build it, you run it" model. These projects are fun in the early days but tend to suck after a few years when the new tech they introduced is no longer shiny, a few people have left, and nobody really knows how it all works anymore.

If you are a bigco and you don't provide a PaaS of some sort, every team will build their own slightly different snowflake - even if they are trying to stick to cloud provider primitives. This eventually gets expensive and hard to support.

It is the eternal battle between centralization and federation of power / decision making / politics, and it will never go away.

So, long live platform teams? I mean, his suggestion for guard rails and DX tooling is literally my definition of a platform team.

All 39 tweets on a single page: https://threadreaderapp.com/thread/1534489514818686976.html

Seems a bit ranty and hand-wavey.

Example:

> Reason 3 for why platforms hurt the business: everything is a nail if you only have a hammer.

> Have a question? Submit a ticket, platform team will get back to you in 3 business days. And their answer will basically be "huh, weird. We will investigate for 2 days, send a ticket to AWS to ask them WTF and play Chinese whispers/Telephone game".

> The platform team is a blocker

Does anyone take this seriously?