Re: [rtcweb] PseudoTCP implementation (Re: realiable data service)

[Cullen Jennings <[e-mail]>]

agree on the high level API you want something datagram oriented... I was just thinking about the congestion control / reliability approach of TCP and fine if a reliable , in order, datagram is put on top of that. 

inline ...

On Jul 23, 2011, at 11:26 , Emil Ivov wrote:

> На 23.07.11 04:33, Cullen Jennings написа:
>> It would be really interesting to see someone write a draft that subsets TCP to make it 
>> 
>> 1) still be congestion safe and TCP friendly 
>> 
>> 2) be valid TCP from point of view nats and firewalls
> 
> Why does it need to be valid TCP? What NATs and firewalls will see is
> valid UDP which is the whole point, isn't it?

I should not have put nats you are right - I'm so used to typing nats and fws :-) On the firewalls that do protocol inspection, they may very well want to be able to parse this layer and even the things inside it. I suspect that code that is there is already valid TCP from the point of view of firewalls so I doubt this requirement really changes anything one way or the other. 

> 
> The fact that the protocol carried in the datagrams is TCP-like only
> matters to the application.

Sort of, lots of firewalls are configured not to allow UDP packets to pass unless they understand what is in them.

> 
> Emil
> 
> --
> http://jitsi.org
> 


Cullen Jennings
For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/index.html