You have way too much faith in software.
MRTG isn't magic, if it's missing a sample point, it displays the last
value one. If sample intervals are missed or randomly scattered due
to MRTG box performance issues, it will wrap 32-bit counters and not
warn you about it.
The rate indicators are primarily for people, so you can see at a
quick glance what's going on, if you're concern about pushing a
limit, what you need to know isn't the average, but the peak and
cross-check with the number of drops...
George
> Subject: Re: [nsp] bit rate accuracy?
> From: Nicholas Bastin <nbastin@opnet.com>
>
> MRTG just polls the various MIB-2 ifOctets (in/out) counters. Unless
> there is a bug in your IOS, these counters don't miss any bytes, and
> thus it is impossible for MRTG to 'miss' any data (unless it loses the
> SNMP query, of course, but that would report 0%).
>
> Anyhow, as to the original question, I've never really been able to make
> any sense of the 5 minute input/output counters on interfaces...the
> numbers never seemed to make any sense, at least compared to any way
> that *I* would compute that number. What I *do* know is that MRTG
> builds graphs that look almost exactly like what my traffic generator is
> showing me, so I really don't care what the interface counters say.
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:23 EDT