Subject: Re: FlowScan question
From: Alexander Kunz (alexander.kunz@nextra.de)
Date: Fri Sep 15 2000 - 02:50:08 CDT
Martin,
----- Original Message -----
From: "Martin Paquet" <mapaquet@videotron.net>
> Sorry, Dave is right, some of you may be have the same performance
> concern...
>
> maybe a distributed environnement could be an issue...but i have not yet
> figure out how to synchronize with rrd files ...
>
>
the solution to synchronize with rrd files is to have a ticket-system
between all processes.
I have running a havy patched flowscan on DualPIII/700. Cflowd is running on
a Sun and the flows exportet to a dir that is shared via NFS. Every flowscan
procces (startet from cron, no loop in flowscan) locks a flowfile from top
and take a ticket-number from a file named "aktiv". when the scan is done,
it looks in a second file "done". this file holds the number from the last
finished process. When done=ticket-1 we are ready to write the rrd, if not
wait in a loop.
So you can start two or more flowscans on a DualPentium or on multiple
server with the flows and rrd's are mounted via nfs.
With two CPU it has the power to scan 1.000.000 flows in "realtime".
When you are interested on the edited flowscan I will send it to you. Be
warned, I'm a fresh perl-hacker but the code ist running for weeks on a
network with 15 flowexporters and havy load.
alexander
--
-- http://www.nextra.de -- INTERNET@WORK -- alexander.kunz@nextra.de --
| Alexander Kunz Backbone Engineer
Nextra communication | Birkenweg 14 D-64293 Darmstadt
service provider GmbH | Tel +49-6151-88008-0 Fax +49-6151-88008-500
-----------------------------------------------------------------------
--
Help mailto:majordomo@net.doit.wisc.edu and say "help" in message body
Unsubscribe mailto:majordomo@net.doit.wisc.edu and say
"unsubscribe flowscan" in message body
Archive http://net.doit.wisc.edu/~plonka/list/flowscan/archive/
This archive was generated by hypermail 2b25 : Fri Sep 15 2000 - 02:50:57 CDT