Question on host/newtork load

[Subhendu Ghosh]

On Sat, 4 Jan 2003, Gilles Poiret wrote:

> Hi, 
> 
> I plan to monitor a large network (spread : worldwide, 30 equipments, 60
> services to begin) I have to have very accurate information, so checks
> will occur every 1 or 2 min, I think.( for instance, to catch a reboot)

An interval of less than 5 minutes is not reasonable when actively 
monitoring a device on the other side of the world.  You basic packet 
latency is going to be in the order of seconds depending on the bandwidth.


> 
> First point
> -----------
> 
> I'm wondering about the type of Nagios host required (CPU, memory,...) in that configuration, and if the volume of equipments/services raise (x2, for instance)
> ( Is a PIII 750 Mhz, 256 Mo RAM enough ?)  
> 
> This host will also run MRTG, to view bandwidth on 20 equipments.

Shouldn't have any problem. - make sure your disk is fast and watch your 
disk I/O

> 
> Second point 
> ------------
> which is the bandwidth overhead generated by those checks ? Is there a way to evaluate it by advance ?
> 

run a couple of the service chesk your are interested in and do a packet 
capture - tcpdump/ethereal...

-sg

> I'd like feedback from people who are using Nagios on that context.
> 
> 
> Thanks for advance,
> 

-- 




-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com