Sunday, May 8, 2011

Network performance testing

I've been busy with teaching and a curriculum development project, and have neglected this blog. Sorry.

About a month ago, I tried a data gathering experiment. I created a simple survey on underground TV viewing in Cuba, hoping that Cubans would reply. Well, the experiment did not work -- a number of people started the survey, but only a couple completed and submitted it. I don't know if there was some sort of mistrust or they just had nothing to say.

I would still like to find ways of getting grass roots data on the Internet in Cuba from Cubans. For example, it would be interesting to get data on international network performance before and after the undersea cable is up and running. That could be done if people at different locations could run timing tests -- from simple pings and traceroutes to more elaborate tests like the Network Diagnostic Tool, which is available on many Internet servers (for example, this one at MIT).

In the mean time, here are a few ping tests run during the evening from the US to three Cuban servers:
  Mean Minimum Maximum St. dev. 634 632 642 2.2 856 707 995 89.9 667 638 885 59.9

If we assume that 250 or 300 milliseconds are due to satellite link latency, these times are still slow. At these speeds, low-bandwidth, asynchronous applications like email would be slow and something like browsing a modern Web site nearly impossible.

Running tests like these from within Cuba would require Web access with the ability to run Java applets. Would anyone be willing to run some tests?

No comments:

Post a Comment

Real Time Analytics