From: Danny.Andaluz@triaton-na.com
Date: Fri Aug 15 2003 - 15:32:35 GMT-3
R1-----R2
R1 (Hub, which has many more PVC's to other spokes)
Access-rate: 1.544
PVC CIR: 256000
R2 (Spoke)
Access-rate: 1.544
PVC CIR: 256000
Hello, Group. I had a situation last night at work that kind of has me
stumped. This does apply to my studies for the lab exam so I thought I
would post it here.
A lot of traffic was being sent to nets off of R2 down the PVC to R2 from
many different sources. This was causing the frame port of the R1(hub) to
become completely saturated outbound causing slowdowns for the rest of the
spokes. I needed something quick to fix this, so I tried configuring the
following on the sub-interface of R1 pointing to R2:
rate-limit output 256000 32000 32000 conform-action transmit exceed-action
drop
Before configuring this, ping response time were at about 700-900 ms. After
configuring, this they dropped to about 80-110 ms along with the frame port
utilization outbound on R1 dropping to about 40% from 100%.
What I was trying to do was keep traffic destined for R2 from using up the
entire frame port of R1. Even if the site itself suffered, at least the
other spokes would be spared. It worked, but not the way I thought it
would.
What resulted was that response times to R2 greatly improved along with
response times to the other spokes. I would think that capping the usage at
256000 would make things worse for R2, but it didn't.
What has me stumped is how the situation for nets off R2 improved?
Thanks,
Danny
This archive was generated by hypermail 2.1.4 : Tue Sep 02 2003 - 18:53:59 GMT-3