Out of curiosity, which side of the dev / ops orga...
# general
p
Out of curiosity, which side of the dev / ops organization does your IDP platform engineering team sit? 🐶 Development (under product development leadership, alongside feature teams) 🐱 Operations / Infrastructure organization (SRE, cloud engineering, etc)
j
We have a platform team org separate from product delivery and SRE/ops. It is split into service, infrastructure, data/ml, and frontend. Among other things, the service platform team owns the IDP, we develop it in-house.
k
I recommend making it separate 🙂 because it fits both: Platform is a Product, which is built with infra/cloud/open source tools
p
Here’s a bit of context: I’m an SRE and over the past 6 months or so have been socializing and building a new team called Release Engineering which is essentially responsible for CD tooling and process including tools like ArgoCD, Rollouts, Crossplane, and applications helm charts, and the automation for release orchestration. I’ve also been leading the effort to define a unified IDP vision and roadmap which includes concerns like dev portal (backstage), ephemeral environments, self service infrastructure with Crossplane, and a generally more “designed” DevX (better dev scripting, etc). About 2 months ago a principal engineer on the product side created a “dev platform” out of the blue with little to no vision or roadmap beyond “improving the testing framework”, which is admittedly a mess, mostly to do with our distributed monolithic architecture. Very recently, engineering leadership has been trying to consolidate front line management so that we have less EMs with few direct reports. They want to combine these two teams (my Release Engineering and Dev Platform). In theory I’m actually okay with this, and think it should have been one team from the start. What I don’t agree with is the placement of the team under product engineering leadership. I don’t believe that this team (which will have 3-5 members) will have the support necessary to maintain so many infra/kubernetes heavy services without be closely linked with the infra/kubernetes teams. Am I wrong?
j
Combined can work, as long as the team has the infra expertise to own that CD tooling stuff still. My company used to have 1 "platform team" but we only split it into separate infrastructure and service platform teams as we grew. (We are at about 700 engineers now)
s
we decided to keep it on the "ops" side, having said that we no longer talk in these terms, there are app eng, data eng and platform eng teams, platform eng is independent and serves both app and data eng as well as secops. I also think there is tons of overlap between platform engineering (current industry definition) and typical devops+sre functions. but this is a huge topic not related to question above.
k
SRE & Classic DevOps is just different Operating Models for organizing teams, tech & expertise. Platform Engineering is not a holy grail for everything ;)