Additional Long Term Testing
Posted by Caleb Phillips on 10 June 2007
In our report, we raised concern about a reduction in throughput experienced at one of our long-term-test sites. We indicated that more research was needed to determine the extent of this reduction. To this end, we carried out another long-term test (at site B - as referenced in the report) from Wed Apr 25 16:28:06 PDT 2007 to Thu May 24 06:05:41 PDT 2007. At this site, and during this test, we found a mean downstream throughput of 1.86 Mbps (for a 5MB file) and 1.66 Mbps (for a 1MB file). This is in line with our results before the reduction at site A. So, at this point we can say that MetroFi has not throttled throughput accross the network, and the reduction at site A must have been either temporary, or, is not 
network-wide. Of course, to explain the cause or extent of the original reduction would require a broader investigation. Other testing metrics are also similar to previous results, implying that MetroFi has not made significant changes to their network in the last month. It is worth noting, however, that we’re still seeing a strong correlation between packet-loss and time-of-day, suggesting that as interference from other networks, or usage, increases, so does packet loss. This new data-set and some graphs are available in the repository.
Leave a Comment
Read more
« Uptown Services report finally appears
Results presentation at PLUG Advanced Topics »