Improving CI/CD Pipeline Speed with Self‑Hosted GitLab Runners and Optimizations

Speed is crucial for developer efficiency in any continuous integration and continuous deployment (CI/CD) platform.

1. Use a self‑hosted GitLab Runner – While GitLab.com provides shared runners that are convenient for quick starts, the biggest single‑step speed gains come from running your own runners on a private cloud where network bandwidth is higher, reducing download and upload times for libraries, dependencies, and Docker images.

2. Cache build dependencies – Storing dependencies in an internal private repository dramatically speeds up builds compared to fetching them from the internet each time. Use Docker images that already contain required dependencies or employ the cache keyword in the pipeline syntax, as well as global caching.

3. Use Alpine Linux for CI images – Choose small Linux distribution images such as Alpine for CI jobs. Larger images like Ubuntu can be 30‑40 times bigger, leading to longer download times. Pre‑download images locally and configure the runner to pull from local storage when possible.

4. Reduce unnecessary job execution – Run jobs only when relevant files change by using the only:changes condition. List all directories or files that, when modified, should trigger the job, including shared dependencies.

Example GitLab CI configuration demonstrating the only:changes usage:

test-example1:
  script:
    - yarn --cwd apps/example1/ test
  only:
    changes:
      - apps/example1/**/*
      - shared-dependencies/**/*

test-example2:
  script:
    - yarn --cwd apps/example2/ test
  only:
    changes:
      - apps/example2/**/*
      - shared-dependencies/**/*