| I respectfully disagree. A platform team will eventually and quick fast become a bottleneck, especially if they offer abstractions over infrastructure, paas, saas etc. Such a team is not any different than previous admin/infra teams. The operation model. When their offering breaks or is incapable, dependent teams wait or is stuck. A team should be self sufficient and be able to decide for themselves how they cope with what they need to move forward in a sustainable manner. That means they should do everything themselves towards enabling there offering. I have worked with so called CI/CD teams at least 10 times and it doesn't foster what devops is all about, which is essentially collaboration, mandate and responsibility for your software, inside a smaller team. It comes down to how you organise and where decisions are placed and carried out. Something that reflects on centuries of command and control organisations where it has been around controlling and not enabling and autonomy. I have worked a few places where the platform decision was taken on a broad spectrum, aws for example, and from there every individual teams could do whatever. Every software/product team should be able to do everything they need to develop,test,deploy,secure and monitor their things. It works in a lot of places so of course it can work in a lot of other places. |