TCP congestion control version: HTCP module vs highspeed module in linux kernel

Question

I found that, in Linux, there are many loadable modules for congestion control algorithms of TCP (cubic, new-reno, veno, vegas,...). But there are two modules that make me confused, one is "HTCP" and the other one is "highspeed". Doesn't HTCP stands for highspeed TCP? So what is the differences between "HTCP" and "highspeed" module here? Thanks in advance for pointing out the differences.

Answer 1

Doesn't HTCP stands for highspeed TCP?

No. It stands for TCP for high-speed and long-distance networks and is described in this document from the Hamilton Institute. HighSpeed TCP is published in RFC 3649.

So what is the differences between "HTCP" and "highspeed" module here?

The common point first, is that they want to turn high bandwidth long distance networks more efficient. The main difference, is that HighSpeed TCP relies on the packet drop rate while H-TCP relies on time elapsed since the last packet drop. As a result, H-TCP seems faster to have its Window Size to recover after a congestion event, which will then give a higher throughput.