← Back to context

Comment by shadowgovt

7 months ago

Not really. The relevant parts of those programs are basically buildable based on a list of their features.

The technology that one would get in an open source situation isn't very usable outside of Google's ecosystem because Google builds software on top of Google's stack. Like, without the monitoring infrastructure they've built or the Borg scaling infrastructure, their software is actually kind of fragile because the ethos is " If it starts to malfunction break it quickly so it can cause a monitoring event and to get replaced by auto restart."

The Google way of doing things is actually not a great way to architect most software that isn't running on a giant data center structure.

That's not something Google invented, it comes from erlang. Systems in erlang (and other beam langs) are designed to fail and die, and get restarted by the supervisory tree.

  • Good observation. I really need to get around to learning erlang.

    It's probably worth mentioning that hypothetically, One could take the source code and port it to third party libraries and kubernetes. But I can't help but think that that would be about as much work as clean rooming it from scratch based on a feature description.