I attended the release party for GitLab 9.0 last week here in San Francisco. It wasn’t oversubscribed, so I got a chance to speak to their CEO a bit about GitLab’s strategy. Apparently though, it’s a topic so readily available they they’ve actually published on it on their site.
One prospect that I was especially interested in was the possibility of an end-to-end development to production pipeline. They obviously host Git already, but have also gone so far as to provide CI services for repositories, and then went even further by providing review apps 1 that are deployed automatically based on Git branches that are pushed to GitLab’s servers.
What if they took it one step beyond by hosting production
apps? Developers could create an account, push code, and
arrange to have it go live at a *.gitlab.com URL from the
same interface that they’re using to merge pull requests.
GitLab does the heavy lifting to make sure that deployments
are continuous, and safe.
Between AWS, GCE, and Azure, the world now has more container services than you can shake a stick at, but they’re all so low level that every serious user ends up either building their own infrastructure management and deployment tooling from scratch, or maintains an open source installation of something like Kubernetes at great personal expense.
There’s still room for a platform that’s not self-hosted and still meets the needs of 90% of users. Even after years of only modest advacements, Heroku is still relevant in this space even though the IaaS goliaths should have surpassed it years ago. They still make getting from zero to production easy, and guarantee a low level of maintenance overhead once you’re there. Docker-based services that ask you spin up and look after your own nodes for them to run on aren’t even close.
GitLab’s potential opportunity here is to own the whole pipeline from code to deployment, a feat totally unparalleled in the world of developer tooling. By baking in metrics, error tracking, and best practices like canary deployments, they could zero in on a level of sophistication that most users would never have built themselves. If done right, it could be akin to Apple’s advantage in providing a superior personal computing experience through total control over their product’s entire stack.
All that said, users right now are expected to bring their own platform, and app deployment isn’t a priority for the company, which is too bad. Competing for business against leviathans like GitHub and Travis isn’t a place that I’d be too excited about getting into, but as demonstrated by 9.0’s feature list, the people at GitLab seem to have a knack for moving quickly on new features despite having a very mature codebase. Hopefully, it’ll keep them in the game.
1 Identical to the Heroku feature of the same name.
Did I make a mistake? Please consider sending a pull request.