I thought I'd put a follow-up to my original question in case someone in the future is looking for more information on this.
I was originally looking for information on this topic because my organization has been having throughput issues with one of our carriers. We have several sites (more than 20), connected with a metro-ethernet type of service. Our main site has a 10Gbps ethernet circuit and one of our remote sites has a 1Gbps ethernet circuit.
In the past, we've had severe packet drop issues going over this circuit. We've spent 8 months troubleshooting this problem, internal to our network and also focusing on the carrier. We recently discovered that our carrier has their edge switch set to an interpacket gap of 64 bits rather than the standard IEEE 96 bits. Without going into too much detail, this causes their switch to transmit ethernet frames slightly faster than our WAN router (Cat 6506) would handle. Since IEEE, and consequently Cisco, puts the IPG at 96 bits, our WAN router interface wouldn't be ready to listen for a frame by the time it was actually receiving a frame from our WAN carrier.
This problem only crept up when transmitting at full wire speed, which we saw during large file transfers between sites.
I still can't find how to change the IPG on Cisco equipment, minus the links I shared above, which cover very specific equipment. Our carrier changed their IPG on their equipment to the IEEE standard of 96 and our problem completely went away. A 4.2 GB file was taking on average over 3 or 4 minutes to transfer across our 1Gbps ethernet circuit. Now with the problem resolved, the transfer takes the expected time of between 20 to 35 seconds.
Even though no one could help, I thought I'd post this follow-up since our carrier, who shall remain un-named in this forum, is a major North American carrier. If our IPG was set to a non-standard time, I'm sure others are affected by this problem as well. I figure this could help someone else out there.
