[Dovecot] 1.0-test21

Moe Wibble t1lt at bk.ru
Mon Jun 21 21:45:17 EEST 2004


On Mon, Jun 21, 2004 at 08:51:55PM +0300, Timo Sirainen wrote:
> On 21.6.2004, at 18:56, Moe Wibble wrote:
> 
> >> - Enabled cache file again. If client asks about something that can 
> >>be
> >>cached for future, it's done. There's currently no smart logic about
> >>when not to do it or when to cache more than was asked for future use.
> >>Currently nothing is ever deleted from cache. Currently it's disabled
> >>if mmap_disable = yes.
> >
> >...does that mean the cache-file will grow indefinitely?
> 
> Yes. Note the "currently" words there :) Will be fixed before 1.0 of 
> course.
> 
> >Actually I find the current maildir-performance quite amazing when 
> >compared
> >to courier or bincimap.  The only other imapd that can hold a candle to
> >dovecot (at least according to my little-bit-out-dated and
> >not-so-representative personal benchmarks) appears to be cyrus.
> 
> I recently saw some benchmarks (measuring system load) comparing 
> Dovecot mbox, maildir and Cyrus. Dovecot was much slower than I 
> thought, Cyrus was many times faster in most tests. Dovecot with mbox 
> was also much faster than with maildir, even though my 0.99.10 mbox 
> code is pretty bad.

Strange...  I'd think that rewriting the mbox files would cause a lot
more performance issues than shuffling around files in a Maildir.
But well, as I mentioned before I'm no friend of mbox anyways and think it
should be completely dropped in favor to more ressources spent on Maildir
performance/features. I do realize that I'm probably in the minority with
that opinion, tho ;)

> 0.99.10 indexes aren't too good, but I still find it a bit strange that 
> Cyrus takes something like 10x less load. I'd think most of it has to 
> do with maildir format itself, that it needs to rename files when flags 
> change, and Dovecot needs to resync the whole maildir after each change 
> in mailbox (and sometimes twice).

I haven't tested dovecot in a high load (multiuser) environment yet
so I can't say much about the actual load in such a situation.
But once the indexes are made (and don't break) what's really
left to cause (unjustified) load?

> I guess we'll need a IMAP-optimized format sometimes soon.

But please not before maildir support is stable again and shared folders
(which I'd claim is a must-have for most corporate deployments) have been
implemented. :)
<rant>
The "proprietary" mailstore-format is the main reason why I'd like
to get rid of cyrus. I'll drop it the day that another imapd provides shared
folders on regular Maildirs with reasonable performance..
Consequently I'd rather like to see dovecot improve on the indexing/caching-
side in order to get most out of Maildir before yet another "prop." format
is invented. I do realize that getting the highest performance out of
server-side searches may require moving to a prop. format. But for me the
drawbacks (uneasy access to the actual mails, probably backup issues/version
incompatibilies) outweight the advantages in most cases. Not to mention all
the implementation effort that could be spent on smarter Maildir indexing
and maybe a separate "search-optimized index" instead ;)
</rant>

Again, that's only my opinion. I'd like to hear others on that...

> Anyway, 1.0 is nearing a state where I'd like to begin hearing 
> benchmarks about it. It's mbox performance should be excellent.

I'm curious too (more about Maildir performance, tho).


greets




More information about the dovecot mailing list