Home | FAQ | Server | Presentations | Mailing Lists/Archives | Member Tools | Links | Sponsors | ContactHome | FAQ | Server | Presentations | Mailing Lists/Archives | Member Tools | Links | Sponsors | ContactI'm sure this is true. Journaling aside, you're comparing Ext2fs with Reiser's b*tree layout.
Ext2fs -> Table based. Rather efficient with low numbers of files, size doesn't matter much, as long as they're not too fragged.
Reiser -> Tree based. A better all around solution for an OS, mostly because an OS has *many* files, and better, the more files you get, and handles frags better, IIRC.
So ... with oracle, which shouldn't use that many files for its db, and typically having entire partitions allocated for nothing other than oracle, yeah. You'd see the good side of the table based exponential function.
on a partially unrelated note, isn't Samba mediocre anyway? aren't we comparing samba to native smb? Does it really have anything to do with linux testing?
I mean ... arent they saying "my applications talk to my applications better than your applications talk to my applications."
and does that suprise *anyone*?
n8
-----Original Message-----
From: Darth BoBo [mailto:EMAIL:PROTECTED]
Sent: Tuesday, May 06, 2003 9:07 AM
To: MLUG Off-Topic Discussion
Subject: Re: [MLUG - DISCUSSION] Windows 2003 66-95% faster the Linux
I've done some reading (research for putting RHAS and Oracle on one of
our big boxes), and there was at least one fairly well written article
that concluded that even though ext3 appeared to be slower in some
circumstances, it was far superior to reiser when actually running Oracle.
The article:
http://afro.d-oh.org/docs/oracle-linux-tuning.html
Rick
Michael wrote:
>I've found Reiser to be quite a bit faster and effecient with space than
>Ext3. I haven't really tested XFS that much.
>
>
>
>>That article frolm VeriTest doesn't do things justice either... it
>>compares Win2003 to RHAS 2.1 with a 2.4.9 derivative kernel that's what,
>>a year and a half old if not more?
>>
>>I think if I went back through the LKML archives, I can find issues with
>>2.4.9.
>>
>>I think VeriTest did a dis-service to its readers. There is an
>>enterprise 2.4.18 based kernel they should have tested with.
>>
>>I'm not suprised with the disk I/O either... ext3 from back then (and in
>>most cases now) is very slow (as is any journaling file system). I'd
>>have liked to see them square off against reiserfs or xfs (for that
>>enterprise feel :-)
>>
>>I'd like to see a 2003 vs. a 2.6 based system or heck, 2003 vs. FreeBSD
>>5.1 (since it'll be out in a month or so).
>>
>>
--
"The last time the French asked for 'more proof' it came marching into Paris under a German flag."
--David Letterman
_______________________________________________
discussion mailing list
EMAIL:PROTECTED
http://mlug.missouri.edu/mailman/listinfo/discussion
_______________________________________________ discussion mailing list EMAIL:PROTECTED http://mlug.missouri.edu/mailman/listinfo/discussion