Arch Linux

0x29a

Member

Registered: 2012-09-28

Posts: 101

Question about partition layout...

Hi there!
I've got some questions about a thing described in
https://wiki.archlinux.org/index.php/Ma … rtitioning

If using a traditional spinning HDD, your partition layout can influence the system's performance. Sectors at the beginning of the drive (closer to the center of the disk) are faster than those at the end.

On the other hand, wikipedia says that HDDs spin with Constant Angular Velocity
Moreover, I can remember from high school (And I'm terrible, terrible at classical mechanics, so please forgive me if I'm wrong!) that points on the edges of a wheel make larger distances in the same time than points near the center of a wheel, so... they move faster?

And yet another thing: Long time ago I've heard from folks @ *BSD forum that hard drive manufacturers design their hardware in a way that the first partition is placed on the drive's edge...

So... How it really looks like?

Thanks in advance for explaining it all!

Edit:

And on the other wiki page, http://en.wikipedia.org/wiki/Zone_bit_recording

On a hard disk using ZBR, the data on the tracks in the outer most zone will have the highest data transfer rate.

Last edited by 0x29a (2014-02-21 19:09:05)

ewaller

Administrator

From: Pasadena, CA

Registered: 2009-07-13

Posts: 19,772

Re: Question about partition layout...

There are quite a few generalities there.  Drive manufacturers can either write more data around the outer edge of the disk, or just have lower data densities out there.  Regardless, the rotational latency will be the same.   What really matters is seek time.  The heads move off the platter when parked.  Seeking to the outer tracks takes much less time than than it takes to get all the way in to the inner tracks.

After every seek, you also incur a rotational latency.  So, if a drive has higher density recording on the outside track, you need to seek less (more data per track) and you do not need to seek as far from the landing zone.

Don't worry about it, you probably won't be able to measure the difference anyway.  At least not in a repeatable way.

Last edited by ewaller (2014-02-21 19:20:19)

---

Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Sometimes it is the people no one can imagine anything of who do the things no one can imagine. -- Alan Turing
---
How to Ask Questions the Smart Way

0x29a

Member

Registered: 2012-09-28

Posts: 101

Re: Question about partition layout...

Thanks for explanations, it's very interesting.

So.. All and all, should not the sentence "(closer to the center of the disk)" be edited out from this article? It's very confusing...
Or. maybe. a more detail-rich description should be written in its place?
I, personally, dont feel competent enough to do that... But it would be nice if someone more experienced decided to dig deeper into this issue
Arch Wiki is a great knowledge mine. I wish it was as perfect as it can get

Trilby

Inspector Parrot

Registered: 2011-11-29

Posts: 29,522

Website

Re: Question about partition layout...

While I would always be in favor of more information in the wiki, I do think it should be suitably organized;  I don't think that page is the right place for a crash course in drive hardware or the physics of how they are read (though such an article could be linked from this page).

The maximizing performace page should provide just the brief pointers on what is needed to put that information into useful practice.  So if it changes, I think the better change would be to remove the ambiguous parenthetical note, and just stick with the general claim that partition layout can affect performance thus X Y or Z partitioning strategies may improve performance[1].  One can always dig deeper: every answer leads to several new questions;  but for practical application most readers are happy to leave the physics to the physicists and just go with the "take-home-message" of what they should do to keep their system running.

[1] Whether this is really true and/or worthwhile is a debate I cannot contribute to - but assuming it is, all the linked wiki page needs is the end conclusions of which partitioning schemes provide which benefits.

Last edited by Trilby (2014-02-22 14:41:19)

---

"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" -  Richard Stallman

0x29a

Member

Registered: 2012-09-28

Posts: 101

Re: Question about partition layout...

Yep. That's exactly what I'm calling for ;-)