|
|
Subscribe / Log in / New account

Runtime filesystem consistency checking

Runtime filesystem consistency checking

Posted Apr 5, 2012 10:00 UTC (Thu) by khim (subscriber, #9252)
In reply to: Runtime filesystem consistency checking by misiu_mp
Parent article: Runtime filesystem consistency checking

More plates means more heads, with possibility for concurrency - that should increase sequential transfer speed.

Good idea. Sadly it's about ten years too late. Today's tracks are too small: when the head is on a track on one plate all other heads are not on this same track. In fact they are not on track at all. They just randomly drift between 2-3 tracks adjacent to each other. That's why you can only use one head actively (how can we use even one if it's all is so unstable? well, it's easy: there are active scheme which dynamically moves head to keep it on track).

If data is written cylinder-wise, the latency should be similar to one-plate disk.

Latency of seeks - yes, number of tracks - no. If you use the same plates then filesystem on a single plate HDD will be roughly five times faster then filesystem on five plates HDD.

That is the main reason we don't see that many of them.

The main reason we don't see many of them is cost. They are more expensive to produce and since they are less reliable they incur more warranty overhead. They are also slower, but this secondary problem.


to post comments

Runtime filesystem consistency checking

Posted Apr 13, 2012 8:47 UTC (Fri) by ekj (guest, #1524) [Link]

So, we "only" need to make the arms move independently then. :-)


Copyright © 2025, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds