Hacker News Reader

That is the peak we see right now at ~20-30% utilization overall network/cpu wise, it can do much more easily.

> Binary logging is a must, because even something like string formatting is simply way too slow, by an order of magnitude.

true, string processing takes most of the resources.

View on HN

Timestamps are even worse. They're slow to get and have a multitude of other problems, like timezones, system clock adjustments (manual or NTP), rate drift, offset errors, etc.

Some solve it by caching a low res timestamp, but that means losing ability to get side channel information about precisely how long time passed between two events. High resolution (micro/nanosecond range) timestamps give so much more information than just time.

We haven't dealt with those yet. But in general we are always pushing everyone to log judicially, use metrics wherever possible (numbers are still easier to manage than strings), use sentry for exceptions logging and introducing zipkin for tracing.

Reply on news.ycombinator.com