Solid-state storage devices and the block layer

Posted Oct 5, 2010 10:24 UTC (Tue) by mjthayer (guest, #39183)
Parent article: Solid-state storage devices and the block layer

It is rather nice that most of the performance work looks like undoing tricks to make rotational media work faster.

Solid-state storage devices and the block layer

Posted Oct 5, 2010 10:49 UTC (Tue) by hmh (subscriber, #3838) [Link] (3 responses)

Unless you're the owner of rotational media, and the optimizations are being permanently undone (instead of being just disabled for non-rotational media).

Solid-state storage devices and the block layer

Posted Oct 5, 2010 10:57 UTC (Tue) by mjthayer (guest, #39183) [Link] (2 responses)

> Unless you're the owner of rotational media, and the optimizations are being permanently undone (instead of being just disabled for non-rotational media).
Is that really the case? I have trouble imagining that they are willing to drop support for rotational media quite this fast.

Solid-state storage devices and the block layer

Posted Oct 5, 2010 11:13 UTC (Tue) by hmh (subscriber, #3838) [Link]

I don't know. I sure hope it isn't the case, we will need fast, optimized support for rotational media for a few years yet...

Solid-state storage devices and the block layer

Posted Oct 5, 2010 11:15 UTC (Tue) by axboe (subscriber, #904) [Link]

Optimizations for rotating media are not dropped. Some of the early SSD work was centered around detecting them properly, that gives you a way to make informed decisions on these optimizations.