Real Time Ascend Maling List Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: (ASCEND) MRTG + Ascend = the age old questions
If you want just the gross statistics over longer periods of time, you could use
the
RADIUS attributes of Session-Time and in and out octets to get an overall
throughput measure. It won't come close to what MRTG shows though.
One thing you could do is to monitor the interfaces on the remote pipeline.
Granted that your SNMP packets will cause additional activity on the line, once
every 5 minutes, but if the customer has a static IP or subnet, that
wouldn't be so bad. This would probably work out fine for nailed links. For
dial on demand links, you'd want to set up an Ascend-Call-Filter in the radius
profile to specify "drop"ing the udp ports 161. The call-filter won't actually
filter -out- the packets, but your side of the connection (and the pipeline too
if you put call filters on both ends) will not reset the Idle timer for any SNMP
packets. So, a connection with only SNMP activity will be considered idle until
a non-SNMP packet passes through.
It's a hack, but if you're in a pinch...
-Will
-----Original Message-----
From: Scott Kozicki <scott.kozicki@bluestar.net>
To: charnock@fastlane.net <charnock@fastlane.net>
Cc: ascend-users@max.bungi.com <ascend-users@max.bungi.com>
Date: Thursday, March 25, 1999 6:20 PM
Subject: Re: (ASCEND) MRTG + Ascend = the age old questions
>"William R. Charnock" wrote:
>>
>> On a pipeline, you can graph the wanidle interface without having to worry
>> about which interface they are actually connected on specifically.
>
>I looked at that, but a) the wanidle interface defaults to 0 traffic and
>b) it still moves around between reboots.
>
>Some people had recommended summing interfaces, but that doesn't give me
>an accurate reading of the bits passing over the pipe (especially if
>they're config'd for multicast, which they are)
>
>I remember hearing that Ascend had a fix for 'roaming' interface
>numbers... something like you could lock interfaces down at an SNMP
>level, or use an OID to walk the int table and pull out the appropriate
>one by name (ie0). Anyone know anything about this?
>
>Thanks again in advance,
>--
>Scott Kozicki
>BlueStar Communications
>
> "5 years in the Internet world is equivalent
> to between now and when the sun burns out."
++ Ascend Users Mailing List ++
To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd: <http://www.nealis.net/ascend/faq>