Observations from Outside Meta

This article is a collection of observations from the point of view of an ex-Meta (formerly Facebook) engineer of 14 years who is now working in other companies.

It will evolve over time as I work in different places and observe how things are done differently.

The good

• Fast compilation time: Where I am now, compilation times are very fast. At Meta, building binaries often took a long time — 20–30 minutes from cold cache. Here it’s more like 1–2 minutes. This is a huge productivity boost. Monorepo and huge C++ dependencies really slow things down and produce multi-gigabyte statically linked binaries. That made building OS images and ramdisks painful and slowed down provisioning.

• Open Source Ecosystem: It’s refreshing to work with widely used open source solutions instead of internal tools. Being in a closed ecosystem for a long time is not good for your career. Many internal tools have external counterparts, but using projects like Kubernetes, ArgoCD, Grafana, Victoria Metrics, and NetBox means I can:

  • Find solutions and troubleshooting tips via public documentation and community forums.
  • Contribute back to the community and share knowledge openly.
  • Apply my learning across different companies and contexts.
  • Leverage the collective experience of the industry instead of being limited to internal knowledge.

  While Meta’s internal tools were powerful and purpose-built, working with open source tools makes knowledge more portable and collaborative.

• I am enjoying Slack but I miss Workplace groups: Believe it or not I am enjoying Slack — it’s more responsive than Workplace Messenger and the integrations are great. I do miss Workplace groups, though. I feel like slack is too synchronous; I should try to treat it more like an async tool and use features such as saving threads for later.

The bad

• I miss stacked diffs, the internal review flows, the internal Meta hg fork. I’m doing okay using jj, but GitLab has no official support for stacked merge requests. There is an experimental glab stack CLI I need to use. The diff reviewing experience is not on par with Phabricator.

• I miss monorepo: Finding and searching code is harder across multiple repositories. I also miss being able to make cross-repo changes in a single commit and the ability to perform large-scale refactors across the codebase.

• YAML everywhere!: YAML on k8s manifests, YAML on CI/CD pipelines, YAML on config files, YAML in Helm charts, YAML in Ansible. I miss configerator (aka configuration as code).

• You miss a lot of internal tools, but the xfb2xfb github repo helps!

The same

• OS/Images build times: Building images is still slow, just as it was at Meta.
• Timezone differences: I don’t think I can get rid of that, it’s just the nature of the work… Still need to work on raising awareness about TZ differences and TZ etiquette.