Hi Didier, Its something totally new! One quick question:
What happened to original ROHC-over-Ethernet code that I had asked for review and submission? Will it be part of original ROHC library or will it be merged with ROHC-over-UDP app in ROHC library? It will take me some time before understanding this new application and its behavior before reviewing it. regards, Raman On Mon, Mar 11, 2013 at 6:07 PM, Didier Barvaux < [email protected]> wrote: > Hi Raman, hi everybody! > > > > > Please share your thoughts on this with me, to take it further. > > > > Sorry for being silent for many days. I had many tasks on other topics > > to complete... > > > > I understand the need for a complete tunnel application using the ROHC > > protocol. That would be great! > > > > About the organisation, I prefer a separate repository for the > > software. That way, the library and the tunnel application could be > > developed at different paces. > > > > I'll come back with some news on this topic next week. Stay tuned. > > Here are the news promised! :) > > > I'm proud to announce a new application. It's a tunnel application > similar the ROHC/UDP tunnel of the ROHC library, but designed for > production! > > > === Origins === > > The tunnel application was designed and developed by Viveris > Technologies [2], my employer. > > Viveris Technologies is a French company working in the IT field. The > company provides to the ROHC project the skills of its engineering teams > in the telecommunications, network and Linux fields. > > Viveris Technologies proposes the knowledge and expertise gained on the > ROHC library to companies and people that are interested in network > header compression, but do not have the skills or the time to integrate > the ROHC mechanisms in their application or infrastructure. > > Kudos to Alexandre Chataignon [3] that wrote the code at Viveris, and to > my boss for publishing the code under the GPLv2+ license! > > > === Features === > > The application is based on ROHC, of course. It encapsulates the ROHC > packets in IP packets (to avoid the UDP overhead). Because of that it > is named IP/ROHC (iprohc in code). > > In addition, we implemented frame packing: several ROHC packets are put > together in one single IP header to reduce the tunnel overhead even > more [4]. > > > === Software architecture === > > The IP/ROHC application is divided into two parts: the server and the > client. The server part handles sessions with several clients > simultaneously. > > The client and the server establish two communication channels between > them: the control channel that uses TCP for reliability, and the data > channel that uses UDP for efficiency. > > The client establishes and closes a session on the control channel. The > client transmits compressed data on the data channel. The control > channel is authenticated and encrypted with TLS. > > Both the client and the server uses TUN interfaces to fake network > interfaces. > > > === Supported platforms === > > The application was tested under Linux, and more especially Debian > Squeeze, CentOS 5, Arch, and Gentoo. It should however work on every > Linux distributions. If not, report a bug and/or send a patch. > > The application requires libyaml, gnutls, and ROHC. A RPM spec and a > Debian package are available in the source tree. > > > === Code / development === > > The IP/ROHC application is hosted on the same project as the ROHC > library. The website, the wiki, the bugtracker, and the mailing list > will shared among the two code bases. > > The code of the tunnel application is however not located within ROHC > library. There is a dedicated Bazaar repository for it [1]. This way, > the 2 applications will evolve at their own rate. > > There is no release of the IP/ROHC application for the moment, however > be sure that code is ready for production: customers are using it :) > > > Regards, > > Didier Barvaux, on behalf of Viveris Technologies > > > [1] IP/ROHC code: https://code.launchpad.net/~didier-barvaux/rohc/iprohc > [2] Viveris Technologies: http://www.viveris.fr/ (sorry, in French only) > [3] Alexandre Chataignon: https://launchpad.net/~xouillet > [4] IP encapsulation means less overhead, but it also means that NAT > gateways won't handle it well in many cases. > > _______________________________________________ > Mailing list: https://launchpad.net/~rohc > Post to : [email protected] > Unsubscribe : https://launchpad.net/~rohc > More help : https://help.launchpad.net/ListHelp >
_______________________________________________ Mailing list: https://launchpad.net/~rohc Post to : [email protected] Unsubscribe : https://launchpad.net/~rohc More help : https://help.launchpad.net/ListHelp
