On any sort of internet connection, speed is of course important.  The fastest response will be direct connections when the two computers are physically connected.  Of course the internet enables connections over thousands of miles but obviously this involves many more hops in the route.   If you start to use proxy servers or VPNs then you add an additional hop in the route which will almost always slow down your connection even more.

Overall speed is obviously one issue, but depending on what you’re doing online then there’s another that may be even more important.  Latency can actually cause a real problem with all sorts of online applications and especially for people playing games online.  If there is a long delay on the connection playing any sort of online action game can be virtually impossible online. If  you don’t believe me try playing Call of Duty using a satellite internet connection!   If you combine these with a slow VPN or even rotating residential proxies then you can seriously impact performance of your link.

TCP Hybla is an experimental TCP enhancement developed with the principal objective of combating the performance decline triggered by the prolonged RTTs typical of satellite links. It consists of a set of procedures that includes, among others:

  • an enhancement of the standard congestion control algorithm (to grant long RTT connections the exact same instantaneous segment transmission rate of a comparatively fast reference connection).
  • the compulsory adoption of the SACK policy.
  • the use of timestamps.
  • the adoption of Hoe’s channel bandwidth estimate.
  • the application and compulsory use of packet spacing methods (also known as “pacing”).
  • TCP Hybla includes only sender-side modification of TCP. As that, it is totally compatible with standard receivers.

For a full description of goals and characteristics of TCP Hybla refer to the publications section.

Performance.
TCP Hybla offers a pretty impressive efficiency improvement to long RTT satellite hookups with respect to TCP NewReno. It may be adopted either as an end-to-end protocol, or as satellite segment transport protocol in PEP (Performance Enhancing Proxy) designs based upon the TCP splitting principle. It can be also used as transport protocol in DTN architectures. See the performance section for further information.

Linux implementation.
Starting from kernel 2.6.13 Hybla has been included in the official Linux kernel. This implementation, based on the “module” Linux technology, does not include the last two Hybla components: Hoe’s channel bandwidth estimate and packet spacing. Their enhancement is mandatory to totally benefit from Hybla performance improvement. To this end, it is enough to patch the official kernel with the MultiTCP package, downloadable from the downloads section.

NS-2 implementation.
A TCP Hybla module has been developed for the widely adopted NS-2 simulation platform. This element can be downloaded from the downloads section.  At the time of writing this has yet to be tested extensively, it should work with all platforms and even with proxies designed for Instagram for instance.

TATPA testbed.
TATPA stands for Testbed for Advanced Transport Protocols and Architecture. It is a testbed developed by Hybla’s publishers to carry out comparative efficiency assessment of new TCP variants (included Hybla) and alternative architectures, such as PEPs (Performance Enhanced Proxy) and also DTN (Delay Tolerant Networks). It could be fully managed by remote through a powerful web interface. For further information see the TATPA testbed and the publications sections.

Projects.
TCP Hybla development is supported by the European Satellite Network of Excellence (SatNEx) project.

Leave a Reply

Your email address will not be published. Required fields are marked *