[Kolab-devel] imap search operation costs
bernhard at intevation.de
Thu Jul 15 16:54:27 CEST 2004
[Switching this to kolab-devel as it is about implementation details. ]
On Wednesday 14 July 2004 17:01, Bernhard Reiter wrote:
> On Wednesday 14 July 2004 13:16, Stuart K. Bingë wrote:
> > Firstly, please could you provide some benchmarks or other quantifiable
> > data that identifies SEARCH operations on Cyrus to be considered "slow".
> > I am under the impression that Cyrus caches the headers of messages,
> > meaning searching would be quite quick.
> Martin told me that Cyrus really has not has optimised implementation here.
> (Because they believe they are not a database.)
> Maybe he can contribute some numbers.
Briefly talked to Martin about this.
He says that single searches are fast, but concurrent searches
pose the problem. So benchmarking seems to be difficult.
The problem with concurrent users seems to come up
with a couple users already. If they change the folder contents,
the indexes will get invalid often, so the searches getting to become
more like linear operations.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 2145 bytes
More information about the devel