Hacker News Reader

Comment by xiphias2 | Hacker News Reader

How many Petabytes of data are you working with?

Maybe your company is too small for data structures and cutting every 0.1ms latency count. As an example at Google I needed director level exception to introduce 0.1ms latency to the ads server (which wasn't needed at the end, but the monetary damage for timeouts is real).

Understanding basic data structures and how the computer works is important in these cases, even if you just need it once in 3 years.

View on HN

The rarity of this scenario is not remotely close to once every 3 years, and not everyone is interviewing to work on something so obscenely performance dependent.

Probably >99% of developers will never encounter something like this in their career.

That's why >99% doesn't work@ fang

It's an improbable scenario for most developers in the world but it's normal day to day work for quite a few developers working at large companies.

There's no surprise that people are screening for the roles they are trying to fill.

Absolutely, but that screening is reasonable. The same screening at run of the mill companies isn't.

Fair, but for those unaware, ads infra at Google etc. are a class of their own because it's the starkest place in the pipeline where latency = -$. More like high-speed trading fintech than regular engineering. They don't care anywhere near as much about services (possibly--but possibly not even--google.com).

Ha, 0.1ms of latency is a heavy cost for what I do these days. I've spent most of the last two weeks clawing back 0.05ms from here and there.

Algorithms absolutely matter, and that will be evident from work experience and education.

This feels hard to believe, what with gmail now taking almost a minute to load in a usable way. :(

Reply on news.ycombinator.com