Traffic Shaping with Frame-relay

[david-schroeder_2]

I have a frame circuit that has a CIR of 256 and an access-rate of 512. The telco has physically provisioned this at the remote site to use all 24 channels and they choke it down to 256/512 in their equipment. My frame-cloud is a T1 at my host site. This works fine and I do not drop any packets when using this.

Recenty, we converted our T1 cloud to a DS3 cloud. We now have a DS3 connection on the host side with the same setup on the remote side.

Under the new configuration, I drop packets at this location. I do not drop packets at any location that used the appropriate number of channels to equal the AR. Currently I do not issue traffic shaping nor did I use it on my old T1 backbone.

My telco is telling me I must use traffic shaping to fix the packet loss problem. What I don't understand is I did not drop packets on this circuit on the old T1 backbone and I do drop packets on the new DS3 backbone for this location.

Is it possible that the telco cannot configure their equipment to handle the situation described above?

Any information would be greatly appreciated.

Thanks,

Dave

[scottmac]

Hi Dave.

The scenario you describe is probably not too unusual. As the pipes start to fill up, newer accounts may have policing enabled.

Your old T1 probably was not policed, but since you changed / upgraded your service, they may have added policing as part of the plan. I believe the SLA / contract should address policing / traffic shaping , etc.

Usually you'll see the packet drop issues sendingg from the fat pipe to the smaller pipe. Generally speaking, you don't want to send more than the receiving DLCI can accept ... or the latency climbs, and ultimately packets get dropped.

Good Luck

Scott

[david-schroeder_2]

Scott,

Thanks for you feedback.

Dave