Hi Wim,

> 40,000 tuples isn't a crazy amount of data. My suggestion is to check your
indexes. Use
> EXPLAIN ANALYZE <sql query> to see if the INSERT/UDPATEs are actually
using the index...

Well, a bit mroe info, I stopped nfacctd earlier, and have just ran a query
on the database, lightning fast. It appears that when it's interting hte
rows, that the point at which everything seems to be slow. I'll be doing
further investigation tomorrow.

> What sort of hardware do you have now?

At the moment it's just an IDE based system, with a Pentium 4 1.8GHz proc,
and 512Mb of memory. Obviously, once I have found the solution that will
work for me, I will upgrade the hardware to something much more considerable
as the number of flows recieved with probably triple/ quadruple once I add
on one of my main transit routers.

My main aim is to find a solution that will enable me to do billing of
customers usages (ie: inbound and outbound bandwidth stats) as well as
produce alerts if certain things start pushing too much data in a certain
period of time.

Cheers,
Steve

Reply via email to