Hacker News Reader

Comment by pessimizer | Hacker News Reader

> and at least a couple more drives for reading tapes.

Why?

Backups are there so that they can be restored. If your only drive is dedicated to writing, then you may never bother reading anything, and that's bad because you should verify your backups.

Also, tape is slow. The MB/s is pretty nice on the latest tech, but a tape is pretty big, so if you have a lot of stuff it'll take a good while. Google says it takes 9.25 hours to write a full LTO8 12TB tape. Which means that if you have a sizable backup, in case of needing a full restore you might well spend a whole week reading tapes.

And that's not accounting for that something might suddenly break, and the time where that becomes important is right when you need something restored urgently.

Even in an "enterprise+++ class" multi petabyte, multi drive, totally integrated from top to bottom tape archive for scientific data, there would be all kinds of errors found by our data validation process that would have failed an archive restore. It's not just cache overruns, some times the tapes or drives just screwed up silently.

If you have 500TB of tape, the chances are that you are reading at least one tape while also needing to write stuff.

I've personally never experienced that scale, I'm sure the industry has some recommended ratio of drives to tapes.

When I evaluated this, it was all about read and write access patterns. So much data coming in for so much amount of time, that needs so much validation, and will be restored so many times in the next few years, etc. It's pretty easy if you know your data flows, but when it's a big question mark, you just kind of throw hardware at it and fix the bottlenecks when they come up. We usually wrote more than we read, but we absolutely needed to keep read capacity open.

Reply on news.ycombinator.com