Jump to content

uTP throughput - unintended consequences?


Recommended Posts

I sit on the end of a "Long Fat Pipe", i.e. enough bandwidth, but >250ms RTT to Europe, and >350ms to the US, in good conditions. When attempting high-speed transfers with these kinds of latencies, regular / standard TCP implementations simply fall to bits. UDP, however, does not suffer the same fate.

Around here, we also depend a lot on wireless connections. With most 802.11 setups, it is not uncommon to have significantly higher UDP throughput than TCP. Common numbers with optimal signal on 802.11G is around 27Mb/s UDP vs 15Mb/s TCP.

Initially I was disappointed with uTP's performance. But it has steadily improved with each version, and after the fixes in 3.0 alpha, I think I have enough hard evidence to raise the point: In both the above scenarios, I see megabits/sec worth of gains, both upstream and downstream, with uTP vs TCP. Megabits - significant.

I know that there are still concerns about the overall efficiency of uTP, and I have no doubt that it will still be developed much further, but I have to make the point to the good people involved in the process: You have (maybe unintentionally) produced something that is worth "gold dust" for us "Rest of World" people at the other end of long undersea cables. As well as the people relying on standard WLAN kit for their connectivity. Good work. Please keep it in mind when you decide on the way forward.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...