...
Throughput vs Int. BW |
|
|
|
|---|---|---|---|
|
|
|
|
We need to look at the outliers to see why they do not correlate well. One possibility is that the international bandwidth may be high but the last mile might not. Often the last mile dominates performances. E.g. Pakistan had a backbone of 155Mbits/s (see the South Asia Case Study) which was hardly used, not congested at all. They proudly showed me it. There were 3 International access points with 57Mbps, 33Mbps & 65Mbps all feeding into a single international cable, thus there was plenty of bandwidth but a single point of failure. However when one dug in we found that the end sites had 1-2 Mbits/s for typically hundreds of people and were heavily congested. It is also possible that the International Capacity measure was at a different time to the PingER measurement.