Re: [tcmtf] About the possibility of having a BOF about TCMTF in IETF87

["Jose Saldana" <[e-mail]>]

Hi,

The idea is not improving header compression techniques here. Of course
improvements in header compression are interesting, but there are other
places for doing that. We only try to put together these three "orthogonal
layers":
- Header compression
- Multiplexing
- Tunneling

The question is that when you compress a single flow, you have to do it
hop-by-hop (which implies a lot of processing in every node), or you can add
a tunnel in order to do it end-to-end. So if you have a single flow, you are
reducing overhead by means of header compression, but the tunnel is adding
overhead again.

But there are scenarios where a number of flows share a path, so in that
case you not only compress headers, but also put a number of packets into a
single bundle, and then the tunnel overhead is shared by all the packets.

Our idea is that RFC4170 can be extended to other services. But RFC4170 is
not only header compression but also multiplexing and tunneling. So the same
ROHC profiles for that services will be suitable for TCMTF. However, we only
want to use existing or new header compression methods. We do not intend to
develop new ones here.

In this paper
(http://diec.unizar.es/~jsaldana/personal/ottawa_ICC_2012_in_proc.pdf) we
presented some results of the obtained savings. For VoIP (RFC4170) we could
save up to 65% of the bandwidth. But for new emerging services, like e.g.,
MMORPG games, the saving can be next to 60%
(http://diec.unizar.es/~jsaldana/personal/sergei_SPECTS_2012_in_proc.pdf).
So why not using the same idea in these new services?

Does this answer your question?

Best regards,

Jose

> -----Mensaje original-----
> De: Eggert, Lars [mailto:[e-mail]]
> Enviado el: miércoles, 30 de enero de 2013 18:46
> Para: <[e-mail]>
> CC: [e-mail]; [e-mail]; Gonzalo Camarillo; Martin
Stiemerling
> Asunto: Re: About the possibility of having a BOF about TCMTF in IETF87
> 
> Hi,
> 
> On Jan 30, 2013, at 18:21, Jose Saldana <[e-mail]> wrote:
> > The main idea (see also the Charter proposal below) is to define a
> > method for saving bandwidth in real-time services which use tiny
packets.
> ...
> > In addition, new header compression methods have been defined (ROHC).
> > So widening the scope of RFC4170 in order to consider not only UDP/RTP
> > but also other protocols is seen as convenient.
> 
> why an extension of 4170 and not new ROHC profiles?
> 
> Lars