Re: Proposed data channel API

On Fri, Nov 11, 2011 at 10:36 AM, Randell Jesup <randell-ietf@jesup.org> wrote:
> On 11/11/2011 1:07 PM, Vincent Scheib wrote:

> In unreliable mode, I would strongly disagree with that.  Out-of-order
> packets should be flagged to the application (if by no other means than by
> providing a sequence #, or by the application being responsible for adding
> sequence numbers to its own data packets), but the application should decide
> if they're important or processable.  I've had connections *at work* that
> could get >1% OOO packets on RTP, for *years*, and was running videophone
> calls from this network all the time.  (Some sort of weird
> router-and-bonded-T1 issue the provider never resolved.)
>
> If the app wants to discard them, fine.  I would be *ok* (though mildly
> concerned) with an app asking the system to discard them for it.

A related question is retransmitted packets.

This is actually a slightly more subtle point than it sounds because suppression
of retransmitted packets at layer N implies either (a) dropping some out of
order packets or (b) potentially unbounded memory growth to remember which
packets have been received and which have not. So, if layer N+1 wants
a guarantee of no retransmissions it generally implies that some really out of
order packets get dropped.

-Ekr

Received on Friday, 11 November 2011 18:53:38 UTC