Bandwidth issues

[wlowe]

I have a T1 link between two offices with a Cisco 1921/K9 using Serial0/0/0 connected to the T1 at each end. We have had periodic bandwidth issues with this link for some time now including with previous routers we were using. Every time we call our circuit provider, they test and say it is with our equipment, but the link starts working again. This latest time I look at both routers and they were showing approx 200 K b/s bandwidth through them but the ping rates end to end were around a 1000 ms. When the link is working normally the ping rate is around 5 ms. Incidentally, our provider again reported that they found no problem.
I guess my question is if a T1 is capable of up to 1544 K b/s, how could 200 K b/s be maxing it out to produce that much delay? Am I seeing this wrong or is our provider blowing smoke at me?

[Georg Pauwen]

Hello,

hard to say. Do you see any output drops or errors on the serial interfaces ? Ping rates of 1000ms are excessive and suggest some sort of congestion along the path. Then again, a T1 link these days would be considered ultra low bandwidth. What kind of traffic is going back and forth between both sites ? Do you have any QoS configured ?

[wlowe]

Yes, I do see output drops. And no, I don't have QoS configured. I'm somewhat knowledgeable on Cisco configuration but in no way an expert. In fact, I really wouldn't be at all that surprised to find that some of this may be my own fault. I'm just trying to solve it with what I have to work with, me. This is why I've reached out here.

[paul driver]

Hello

You could try enabling you own testing to obtain a more acurate assessment of this delay with IP SLA

Please review it here

res

paul

[wlowe]

I'm not sure how to do that.

[Joseph W. Doherty]

How are you measuring the 200 Kbps?

In extreme congestion cases, a very high drop rate will force TCP flows to slow such that their transmission rate is very low.  Or, highly delayed ACKs will also slow TCP transmission rate.

Your high ping times of a second second imply some built up queue in your path.

[wlowe]

Joseph, the 200 is taken from the dashboard interface.

[Joseph W. Doherty]

What dashboard? What is its monitoring interval?

[wlowe]

CCP - Cisco Configuration Professional Express 3.1. When I point a browser at the IP Address of my router it gives me a Dashboard of status. The bandwidth listing there says the Bandwidth is 200 Kb/s. That reads for me at least, as 200,000 bits per seconds passing through the router.  Am I wrong?

I also believe that a T1 circuit should be capable of passing up to 1544 Kb/s. by my math what is going through my router is less than 10 percent of that max. How can that drive the ping rate from 4 to 5 ms to well in the thousands?

[Joseph W. Doherty]

(Hmm, thought I responded to this - if you see a similar earlier reply, it and this are here because I don't see it now.)

Ok, good chance your dashboard is showing a five minute usage average.  During a five minute perios you may truly have a low average usage, but you could also have occasional high usage bursts, which won't be "seen" in a long term average and they might account for your occasional high ping times.

Can you configure your routers with their CLI?

If so, add something like:

policy-map CBWFQ
 class class-default
  fair-queue

Serial0/0/0
 service-policy output CBWFQ