Hacker News Reader

Comment by citrin_ru | Hacker News Reader

I would argue a web server which processes all (or most) requests is a performance sensitive part of the system. Straightforward rewriting from Python to Go in many cases gives at least 2x-3x performance improvement which on FAANG scale can mean huge difference in hardware cost.

View on HN

It's probably cached to seven hells, which makes that 2-3x improvement much less relevant.

Wordpress wouldn't be this successful if this approach didn't work. After all, without caching it will happily take over 20 seconds to render the main page.

Developer velocity is extremely important at FAANG scale.

Is it really? Don't they work really slowly compared to most of the industry? I really get the impression that it's slow progress for things at FAANG.

Not really, especially Facebook (Meta) moves really fast. They have a slogan "Move fast, break things", okay not "Break things" anymore. But you get the gist, in my experience working at almost all of FAANG, Meta and Netflix move the fastest.

I think AWS must be the slowest of the bunch tbh. Too many meetings.

I agree, but Google is the slowest.

Why do you say that?

It could be different for other teams, but in my team we spend a lot of time writing design docs, discussing new features (which we never end up implementing), taking a lot of time in code reviews, making sure that public changes are backward compatible. Hence at the end producing very little output and wasting tons of time in the due process.

Sounds like us tbh. So many hours wasted writing deeply technical documents that only few engineers have the background to go through, too many hours wasted on huge ass documents supporting PoCs. Ugh.

Slow compared to startups, very fast compared to similarly sized companies, even other tech companies.

Reply on news.ycombinator.com