From: smaljure@xxxxxxxxxxxxxxxxx
Date: Thu Sep 28 2000 - 11:44:39 GMT-3
Hi
A friend of mine who was doing some research on the SNMP statistics
collected by Cisco routers came up with similar results. He was trying to
measure load on the Cisco interfaces as shown by the "sh int " command. He
basically generated huge amounts of traffic using some test
video-conferencing software and tried to see if the "sh int " command
represents correctly the actual load on the interface.
When the traffic levels were 70% and above, the router pretty much gave up
on showing the correct load statistics on the "sh int " command. We had this
escalated to some people in Cisco and they said that when the router is
heavily overloaded, SNMP processes get very little processor time and hence
*may* not correctly represent statistics. Basically the advice was to move
on to bigger platforms/VIP/etc...
But anyways what we realized is that SNMP/ICMP stats and response times are
not accurate means of measuring the status of an overloaded router.
Sanjay
-----Original Message-----
From: achau@au1.ibm.com [mailto:achau@au1.ibm.com]
Sent: Thursday, September 28, 2000 7:01 AM
To: ccielab@groupstudy.com
Subject: A question about cisco Ping
Hi all,
I have set up a network management software to monitor a cisco router over
a wide area network. I could ping from the router to the NMS and got 99%
successful rate. But I could only receive no more than 95% successful rate
if I pinged from the NMS bak to the router. The ping utility on the NMS is
a bit different from the one on cisco router. Unlike cisco router, the NMS
sent ICMP echo request on a periodic basis (e.g. every 1 second).
Can anyone tell me which ping method is better and more reliable? How come
I got more packet drops using the one with NMS?
Thanks in advance.
Regards,
Andrew
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:25:08 GMT-3