[Tsvwg] Re: [e2e] What's the benefit of out-of-orderprocessi

[David P. Reed]

At 11:09 AM 9/18/2001 -0600, Hilarie Orman wrote:
>One could regard the pointer as a "covert channel" and interpret it
>as data.


One could, but remember what it carries is a non-decreasing sequence 
number, and it is un-ACKed.  Thus, encoding a bitstream in the sequence of 
urgent pointer values is quite hard (putting an urgent pointer value of N 
in a packet at the sender may never be received by the receiver, because 
subsequent retransmissions of that packet may contain higher values of the 
urgent pointer superseding N).

The "channel capacity" of the urgent pointer as a subchannel of TCP is 
certainly non-zero, but highly conditional on the actual channel capacity 
of TCP and the discard rate of the network.


- David
--------------------------------------------
WWW Page: http://www.reed.com/dpr.html