Hi there, I’m the manager of Cloud Functions for Firebase. I wanted to note that Firebase is getting better for large deployments and we’re continuing to invest in the area.

First, the article might have been written before our recent feature to skip deployments of unmodified functions. If the source, env, and secret metadata SHA has not changed, we skip deploying that function. At minimum, this means large deployments that fail can be retried and make progress (more on that below).

Second, we’ve started investing in a feature called “codebases.” At its simplest, codebases are multiple folders of functions. This makes it easier to deploy single codebases, but it also means you’re likely to skip deploys of functions in codebases other than the ones you’re working on. If you want to invoke functions in another codebase, consider a Pub/Sub, Task Queue, or Eventarc Custom Event (coming in a few weeks) function depending on the feature set that suits you best. Codebases are just getting started and we have hopes to develop this feature in the future.

Finally, we’ve set aside budget in Q4 to rewrite part of the core of our deployment logic to improve the way we batch, backoff, and retry function deployments in the face of quotas (yes, they are retried multiple times, even when we give up due to quota errors). We hope to substantially raise the reliability of 100+ function deployments.