We're a small data team (4 devs) building a modern analytics stack on GCP. Our repo is mostly Python (dbt & some custom pipeline code) and Terraform. Right now, we're using GitHub Actions because it's free, but our workflows are getting messy and the 2,000-minute monthly limit feels tight. We need to level up our CI/CD without blowing our limited ops budget.
I'm looking for concrete comparisons on cloud-native CI/CD options for a startup. Our must-haves:
* **Cost:** Predictable, scales with usage. Low/no cost for lightweight pipelines.
* **Managed infra:** We don't have bandwidth to manage an agent fleet.
* **Native with cloud:** We're on GCP, so good integration there is a plus.
* **Speed:** Faster than GitHub Actions on the default runners for medium-sized jobs.
The main contenders I'm evaluating are:
* **GitHub Actions** (with self-hosted runners on GCP?)
* **GitLab CI** (SaaS)
* **CircleCI**
* **Cloud Build** (GCP native)
* **Buildkite**
I'd love to see benchmark numbers or experiences for similar-sized teams. For example, a typical pipeline for us looks like:
1. Lint & unit test (Python, ~3 mins)
2. Integration test (spins up BigQuery dataset, runs dbt, ~8 mins)
3. Deploy (Terraform apply or dbt run on prod, ~5 mins)
Which platform gives the best balance of price, performance, and ease of debugging for this pattern? I'm especially curious about cold-start times and the cost of parallelizing steps.
--diver
Data is the new oil - but it's usually crude.