RE: OCFS is as fast as raw?

Erm, I realized I didn't really address some of the stuff that was brought up in this email.

The oracle fileutils are to allow normal utilities to access those files, since OCFS only supports direct i/o (non-buffered i/o), and none of the normal fileutils/textutils call O_DIRECT when opening a file (since they are not aware, really, of the concept of direct i/o).

Async i/o should be a performance boost - were you using the 1.0.9 release? I'm about 80% sure the 1.0.8 and earlier releases don't support async i/o.

And the comm_voting enabled should be a performance boost - that enables network lock management (rather than the disk voting used otherwise). Another big performance optimization is to maximize your block size - that reduces the lock traffic.

In fairness, I also have to say that we really haven't put it through its paces - that particular cluster got repurposed for other tasks once we heard about ocfs 2.0, and I haven't come back to it. 2.0 will be a much more detailed testing scenario for us.

Thanks,
Matt

--
Matthew Zito
GridApp Systems
Email: mzito_at_gridapp.com
Cell: 646-220-3551
Phone: 212-358-8211 x 359
http://www.gridapp.com



> -----Original Message-----

> From: ml-errors_at_fatcity.com [mailto:ml-errors_at_fatcity.com] On 

> Behalf Of Jesse, Rich

> Sent: Thursday, September 18, 2003 6:25 PM

> To: Multiple recipients of list ORACLE-L

> Subject: OCFS is as fast as raw?

> 

> 

> After perusing the OOW'03 presentations, I see that Win 

> Coekaerts' presentation says "IO throughput [on OCFS] is 

> equivalent to RAW IO".  Has anyone seen this behavior?  It's 

> entirely possible (probable!) that I didn't do something 

> correctly, given the bastardization I needed to do (installed 

> on RH9 because AS2.1 won't support our hardware), but I saw 

> OCFS performance about 40-50 times slower than RAW (rough estimate).

> 

> I didn't implement the OCFS fileutils as I wasn't going to 

> manually do anything with the files.  Surely the oracle 

> binary doesn't call dd, cp, and mv directly for it's file 

> management and data I/O, does it?  Is it the "comm_voting=1" 

> in the /etc/ocfs.conf file?  Or is it more likely to be async 

> IO (my guess)?  Or a combo?

> 

> Anyone with experience on this?  Anyone attend Mr C's presentation?

> 

> Thanks,

> Rich

> 

> Rich Jesse                           System/Database Administrator

> rjesse_at_qtiworld.com                  Quad/Tech Inc, Sussex, WI USA

> -- 

> Please see the official ORACLE-L FAQ: http://www.orafaq.net

> -- 

> Author: Jesse, Rich

>   INET: Rich.Jesse_at_qtiworld.com

> 

> Fat City Network Services    -- 858-538-5051 http://www.fatcity.com

> San Diego, California        -- Mailing list and web hosting services

> ---------------------------------------------------------------------

> To REMOVE yourself from this mailing list, send an E-Mail message

> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') 

> and in the message BODY, include a line containing: UNSUB 

> ORACLE-L (or the name of mailing list you want to be removed 

> from).  You may also send the HELP command for other 

> information (like subscribing).

> 


-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: Matthew Zito
  INET: mzito_at_gridapp.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).