[rezonant]

I read again his argument.

> Keeping all of your application logs and telemetry forever is expensive, and I can't recall a single time when having more than a day's with of history was ever useful in tracking down an operational issue.

That doesn't say don't store any, that says you can get by storing a 24 hour period. And his broader point is that it should be time bound, that storing these metrics indefinitely isn't useful and can be very expensive.

I'm of the mind that a week or two of fast online access is the right amount myself (with offline "cold" storage of logs for a longer period), but the overall premise that storing logs and infrastructure metrics forever is unnecessary and wasteful.