[Xapian-discuss] Search performance issues and profiling/debugging

[Olly Betts]

On Thu, Oct 25, 2007 at 05:06:16PM -0400, Aleph Thomas wrote:
> And exist a way of  use the Xapian DB in a cluster?

You can split a large dataset into several Xapian databases and store
them on separate server, with searches run on each database separately
and the results merged to provide a single MSet for showing to the user.

Actually, it's rather like the map/reduce paradigm which Google have
recently popularised now I think about it!

> The think is that if I
> have a millions of docs I need a very good results of that search, maybe in
> only one PC, I not can get all the good results that I need.

You should be able to search millions of documents with Xapian on a
single PC.  For example, http://search.gmane.org/ is a single server
and the index size is just under 50 million documents.  It does depend
on the query rate though.

Cheers,
    Olly