I really want support for this however with this model there is no way to 
determine if the message begins with an ack frame or a data frame. You could 
determine if a ping had an ack frame by checking the PDU size but not any 
messages that are variable in length.

Julian Cain

On Apr 16, 2010, at 2:22 PM, Michael Chen wrote:

> Hi,
> 
> I want to propose a base draft enhancement for UDP transport:
> 
> When a Reload request and response is exchanged between two and only two
> peers (single hop, no forwarding peer) using DTLS/UDP, the responding
> peer should have the option of sending both the ACK frame for the
> request and the DATA frame of the response (or its first fragment) in a
> single DTLS/UDP packet. Therefore, the combined PDU before DTLS
> encryption will be:
> 
>    [ACK frame]
>    [DATA frame]
>    [RELOAD header]
>    [RELOAD message]
>    [Security block]
> or
>    [ACK frame]
>    [DATA frame]
>    [RELOAD header]
>    [First message fragment]
> 
> The logical order of the combined PDU is identical to that of a TLS/TCP
> response if you take D/TLS and fragmentation out of the picture, so the
> impact on programming complexity is small. The benefits are obvious: 50%
> reduction on network traffics, encryption, decryption, etc.
> 
> A while ago, I proposed another single hop enhancement, which says that
> the security block of a single hop request and response may be left
> empty, because both parties are authenticated on the TLS and DTLS level.
> 
> Both of these single hop enhancements should be considered and
> reconsidered. Please make your opinions known.
> 
> Thank you
> 
> Michael Chen
> 
> 
> 
> 
> 
> 
> 
> _______________________________________________
> P2PSIP mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/p2psip

_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip

Reply via email to