Yonatan Ben-Nes wrote:
> Actually I even started to look on other solutions and maybe you can say
> something about them also.. maybe they can help me:
> 1. Omega (From the Xapian project) - http://www.xapian.org/
You could certainly do this with Xapian and Omega. With only 5
million records it should be very quick.
The easiest approach would be to periodically dump the SQL tables
and build a new Xapian index which reflects the SQL database - you'd
probably want to customise the "dbi2omega" script in the Omega
distribution. This approach works particularly well if the tables
are updated in a batch fashion (one big weekly update, say).
Alternatively you could hook into whatever updates the SQL database
and get it to make corresponding updates to the Xapian index. That
has the advantage that they'll always be in step, but is probably
more work to set up.
The main drawback compared to doing everything in SQL is that you'd
have two systems to deal with rather than just one...
Cheers,
Olly
