[Xapian-discuss] A few questions wrt Xapian

[Charlie Hull]

Olly Betts wrote:
> On 05/11/2008, Henka <henka at cityweb.co.za> wrote:
>> Olly, could you possibly comment?  I'd be happy with one-word terse comments
> 
> I could, but Richard has answered already.
> 
> Apologies for being less responsive than usual - I'm back in the UK for a few
> weeks and currently my internet access is the local library which firewalls ssh.
> 
>> 5.  This is difficult I know:  how does Xapian compare
>> performance-wise?  Has anyone done any basic benchmarking?
> 
> I've not seen many comparative benchmarks (or even qualitative studies) between
> Xapian and other systems and I'm not aware of any that I find very satisfactory.

It would be great to have these though, for persuading people to use 
Xapian. The main hurdle would be cost of any commercial system you're 
benchmarking against (not a problem if comparing with Lucene, Sphinx 
etc. of course).
> 
> A benchmark really needs to have the data used available and include enough
> information to allow a third party to repeat it, and few do.  Often
> people don't even
> mention which versions of the software they are comparing!

Another issue is that every application is different; someone indexing a 
  few hundred huge PDFs will have different requirements to someone 
indexing millions of tiny SQL records.

When we're talking to people about Xapian, we show them systems powered 
by Xapian in the real world that work well: for example gmane, 
TheyWorkForYou or mydeco.

It would be great to expand this list: we have the Users list on 
xapian.org of course, plus the scratchpad list on the wiki, but I'm sure 
there's many more systems out there.

Charlie
> 
> But generally the feedback I've had is that Xapian performs well.  If you do see
> areas where we do worse than other systems, it probably means there's scope
> for profiling to find bottlenecks.
> 
> Cheers,
>     Olly
>