On Wed, Sep 30, 2009 at 10:18 PM, Manvendra Bhangui <mbhangui@xxxxxxxxx> wrote:
The spamassasin example was at Sify. At that point we were hit by spam andThanks, will go through these tomorrow morning.
virus
specifically targeted at Sify. And the servers were just not able to cope up
with the load. One night of frantic coding and the spamassasin was replaced
by bogofilter. A program called qmail-smtpd was modified to look at the exit
code of bogofilter and discard the mail at the entry points.
bogofilter is a very fast bayesian spam filter available at
http://www.bogofilter.org
Currently we are doing some of this stuff at my current job. Just few days
back we have got a huge performance gain by moving to sphinx -
http://www.sphinxsearch.com/ ;(written in C++) and we are
again going to save on servers by moving away from Lucene -
http://lucene.apache.org/java/docs/ ;(written in Java)
Take a look at
http://highscalability.com/
It explains the architecture of high scalable sites like youtube, flickr,
etc. I have personally gained a lot from the above site.