Remove the Serverless product category
Following the changes in direction and admitting that we've made wrong product decisions in the past, I'd propose deprecating the current GitLab Serverless category (again) and remove the relevant features/code.
Pros:
- we don't work on it, but we still have to worry about it
- there are almost no users
- when a user tries it, they leave unsatisfied
- our current approach misses the main value offering of serverless
Cons:
- we wouldn't be able to market it in the future
Serverless
Serverless is both a topic (including AWS Lambda support) and product category (GitLab Serverless). These two should be strictly separated! In this issue, the topic is always called serverless (minor case), our product is always called GitLab Serverless.
My (Viktor) opinion is, that
- the serverless market will grow substantially in the coming years. It's importance won't be as huge as that of Kubernetes, or IaC, but it's still gaining adoption at an increasing rate especially as edge offerings solidify
- GitLab's approach to serverless with GitLab Serverless was a dead-end. At the same time, building strong support for Knative inside of Kubernetes management might be a viable future direction. Similarly, AWS Lambda having Docker support, provides the ability of a platform-agnostic serverless offering. Still, our current offering of far from valuable and misses the market.
As GitLab Serverless is based on Knative, and uses mostly its alpha
and beta
APIs, that are deprecated (and are expected to be removed from Knative in 9 months were already removed), I (Viktor) expect GitLab Serverless to become obsolete sooner or later (it is probably the case by now). As a result, I recommend removing it from our Product Categories altogether, and stop supporting it.
Timeline
Remove our serverless offering in %14.0