| To: | linux-xfs@xxxxxxxxxxx |
|---|---|
| Subject: | Re: Dealing with a very large file -1,5TB |
| From: | KELEMEN Peter <Peter.Kelemen@xxxxxxx> |
| Date: | Thu, 25 Sep 2003 15:37:26 +0200 |
| In-reply-to: | <3466.10.1.200.117.1064496571.squirrel@imap01.ch.sauter-bc.com> |
| Mail-followup-to: | linux-xfs@xxxxxxxxxxx |
| Organization: | CERN European Laboratory for Particle Physics, Switzerland |
| References: | <3F72E2D4.1030109@kth.se> <3466.10.1.200.117.1064496571.squirrel@imap01.ch.sauter-bc.com> |
| Sender: | linux-xfs-bounce@xxxxxxxxxxx |
| User-agent: | Mutt/1.5.4i |
* Simon Matter (simon.matter@xxxxxxxxxxxxxxxx) [20030925 15:29]:
> Did you check the raw write speed of your devices using dd? When
> I did tests in the past I got almost the same speed with XFS
> like raw. Until you are sure your devices perform much better,
> XFS should not be involved too much.
What we observe here is that dd(8) write speed is significantly
lower than XFS write speed, for large streaming writes (32x or 64x
RAM size) on big RAID devices. While XFS churns away with 150M/s,
dd(8) struggles with 55-60M/sec (regardless of blocksize).
Peter
--
.+'''+. .+'''+. .+'''+. .+'''+. .+''
Kelemen Péter / \ / \ Peter.Kelemen@xxxxxxx
.+' `+...+' `+...+' `+...+' `+...+'
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: Dealing with a very large file -1,5TB, Simon Matter |
|---|---|
| Next by Date: | Re: Dealing with a very large file -1,5TB, Christoph Klocker |
| Previous by Thread: | Re: Dealing with a very large file -1,5TB, Simon Matter |
| Next by Thread: | Re: Dealing with a very large file -1,5TB, Christoph Klocker |
| Indexes: | [Date] [Thread] [Top] [All Lists] |