. It is quite fast. I've used it 3 years without a glitch. With Suse8 though I had to use ext3, as Reiser does not yet support filesizes that I need now. BTW, journalling not only is faster at crash recovery, it is more reliable. It has a running journal of all transactions and does them as it can. If the system crashes it doesn't have to go out looking for errors and hope it finds them... it knows exactly what has/not been done. Ext3 does seem slower to me. As soon as the new Reiser is out of beta I'll switch. BTW, Suse8 seems so smooth and steady. What a relief. Only real problem I have is in Konq when you try to edit a file, "KDEInit could not launch 'kwrite'". Also I noticed in file types, must apps are double-entered. May be a related setup problem. Nothing on Google. On Monday 29 April 2002 13:11, Ewan Leith wrote:

Reiserfs uses a very different filesystem architecture from ext2 and other filesystems, which makes it much much quicker at handling small files (its performance on large files is good, but not amazing).

If you want some specific information on reiserfs, check out www.namesys.com, the webpage of the developers.

Ewan

On Mon, 2002-04-29 at 18:45, Manral wrote:

...

Dear friends Could some one explain as to how RaiserFS is faster than ext2 file system ... I am setting up an email server should i make only the /var partition raiserfs or should i make all the /boot, / ansd /usr etc all of them under raiserfs .

Can some one explain the improvement in performance in cas i use raiserfs ..how does it improve the performance ... regards neeraj ----

-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com

-- BLISS is ignorance

On 29 Apr 2002 19:11:13 +0100 Ewan Leith wrote:

Reiserfs uses a very different filesystem architecture from ext2 and other filesystems, which makes it much much quicker at handling small files (its performance on large files is good, but not amazing).

I hate to spoil your party, but my experience does not bear this out. I use sylpheed for a mail client. It uses MH style mail handling, in which every individual email in a given folder is a separate file. Hence, on high volume mailing lists like suse-linux-e, for example, many, many small files. Thousands if I'm not diligent about purging old ones. For a while, I had an elaborate archive system set up to keep old messages, sorted by month, because once the number of messages gets up to 10k or so, performance really starts to drag as far as opening and closing folders. The normal 3-4k/month level allows sylpheed to zip right along, w/ only a couple second delay opening or closing a folder w/ 3k+ messages in it. This is on a Celeron 266 w/ 128MB RAM, IDE HD, running XFCE for a desktop to conserve resources. When I first started using sylpheed, I was running RedHat 7.1 or 7.2, on an ext3 /home. I switched back to SuSE 7.3, and changed to reiserfs. Major performance **HIT**. To put it in perspective, where before I was able to run XFCE as the WM, galeon w/ about 8-10 tabs open, sylpheed, pan, The Gimp, several rxvt windows, one running dnetc (distributed.net client) and another running top, the others running ssh connections to other machines, w/ no performance problems (other than it being a Celeron 266 :( ), now sylpheed ran slower when it was **the only application** running. I am not alone in this observation. I thought I was on drugs, had done something wrong setting up my system, something. Then I noticed on the sylpheed mailing list, that the creator/maintainer of sylpheed had noticed the same thing. His solution at the time was to stop using ReiserFS for the directory where his folders were stored, and go back to ext2/3. I changed back, and noticed a jump in performance. Right now, my machine has ReiserFS for the local hard drive, but the small LAN server that I use has /home exported, and it is formatted w/ ext3. Perhaps it is sylpheed's fault, though I kind of doubt it. It uses a mail system (MH) that is intentionally designed to deal w/ lots of small files. I just tire of seeing people quoting that ReiserFS is supposed to be so great at handling lots of small files, and saying that such-n-such 'benchmark' says so. Theory-to-practice says otherwise, as far as I'm concerned. YMMV, Monte -- All right, breaks over. Back on your heads!!

Alle 20:29, lunedì 29 aprile 2002, Jerry Feldman ha scritto:

Good question I would like answered. I asked a question either with SuSE or with my LUG about the relative performance of the various jounal file systems, and all paid a performance penalty. The only performance plus of the journaling file system was in terms of crash recovery where the fsck is much faster with a journal than without.

That is not true in my experience. I have 5 gigabyte free in my hard drive so I have been doing some filesystem benchmarking with bonnie++. ReiserFS is the fastest, XFS follows with a small gap. Ext3 is slower than Ext2, and both are not too far from ReiserFS and XFS. JFS looks slower. I have been testing it with 500m file, as bonnie++ advices. Praise

Alle 21:27, lunedì 29 aprile 2002, Herman L. Knief ha scritto:

->That is not true in my experience. I have 5 gigabyte free in my hard drive so ->I have been doing some filesystem benchmarking with bonnie++. ->ReiserFS is the fastest, XFS follows with a small gap. ->Ext3 is slower than Ext2, and both are not too far from ReiserFS and XFS. JFS ->looks slower. ->I have been testing it with 500m file, as bonnie++ advices.

How much memory do you have in your system?

256M. Bonnie++ advises to use at least double size than RAM. Praise