Why gittan is SaaS only

We get asked about self-hosting regularly. The answer is no. Not because it is technically impossible, but because it is not how this product works best — for you or for us.

Shared infrastructure is cheaper

gittan's pipeline speed depends on warm runners and pre-pulled images. Runner pods are standing by before you push. Common base images are cached on every node. This infrastructure is expensive to run — and it is economical because hundreds of customers share it.

When a runner finishes your pipeline, it is immediately available for another customer's push. When we pre-pull an image, every customer that uses it benefits. Node scaling handles bursty workloads across all customers, not just one. The utilization is high because the load is diverse.

A self-hosted gittan would need the same infrastructure, but serving only your organization. You would pay for warm runners that sit idle most of the day. You would pay for pre-pulled images that only your team uses. The economics do not work unless you are running pipelines at a scale where you would be better served by a dedicated conversation with us about your needs.

Many moving parts

gittan is not a single application you install. It is a distributed system: Forgejo for git storage, ScyllaDB for data, NATS for messaging, a runner fleet for pipeline execution, a DAG executor for step orchestration, an API server, and a web dashboard. These components need to be deployed, monitored, upgraded, and scaled — together.

We could package this in a Helm chart. But installing it is not the same as operating it. When ScyllaDB needs compaction tuning, when NATS hits a slow consumer, when the runner fleet needs rebalancing after a node failure — that is operational knowledge specific to this system. Your platform team would have to learn it. Ours already knows it.

Self-hosting a distributed system means your organization takes on the operational burden of a system they did not build. Some organizations want that control. For most, it is overhead that distracts from the thing they actually want: fast, reliable CI.

It is our service

We built gittan. We operate it. We are responsible for its uptime, its performance, and its reliability. That is the deal. You pay us to run it well so that you can focus on your software instead of your tools.

This is a deliberate choice, not a limitation. We want to know exactly what every customer is running. When we ship an update, it reaches everyone. When we find a bug, we fix it once. When we optimize pipeline scheduling, every customer gets faster. There is no version fragmentation, no "works on the latest release but you are three versions behind."

Data residency

The most legitimate reason for self-hosting is data residency. gittan runs in EU infrastructure. Your source code, pipeline logs, and metadata stay in the EU. If you need a specific region or dedicated tenancy, we are open to that conversation — but it is still our infrastructure, managed by us.

If you need self-hosted git

If your organization requires self-hosted git hosting, Forgejo is excellent. We run it under the hood ourselves — we know it is good software. What it does not have is gated push with fast pipelines, team metrics, or team-level reports. Those are the things gittan adds, and they work best on infrastructure we operate.