Hi QUIC wg,

I would like to share a draft 
https://datatracker.ietf.org/doc/draft-shi-quic-structured-connection-id/[1] 
and hear your opinion. Inspired by the use case outlined in 
https://datatracker.ietf.org/doc/draft-kaippallimalil-tsvwg-media-hdr-wireless/[2],
 my proposal empowers hosts to convey crucial media metadata (such as I frame/P 
frame/Frame size/Burst/Delay requirements) to the network using QUIC Connection 
ID. This proactive approach aims to optimize resource management, thereby 
maximizing network utilization and enhancing application performance. 

Building upon the existing architecture of the 
https://datatracker.ietf.org/doc/html/draft-ietf-quic-load-balancers[3], my 
proposal expands the server ID to incorporate a metadata field within the 
connection ID. Since the metadata may vary between different types of 
applications and networks, the metadata template is introduced.

You may find similarities between my proposal and 
https://datatracker.ietf.org/doc/draft-zmlk-quic-te/[4], which was presented 
during IETF 117 and faced certain critiques. To address potential concerns, 
I've engaged in private discussions with several people and iterated on the 
draft accordingly. Here's how my proposal tackles some key issues:

1.      Why not DSCP? Because media metadata will vary packet by packet, there 
is not enough space in DSCP.
2.      What about privacy? Learning from past endeavors, the metadata included 
in the connection ID strictly pertains to the characteristics and requirements 
of the application traffic, without divulging a specific user identification. 
By employing a standardized metadata template, transparency is enhanced, and 
the risk of fingerprinting can be minimized, especially if templates are 
governed by an IANA registry(see Option 1 in Section 3).
3.      What about trust? Similar to the encryption employed in the QUIC load 
balancer for server IDs, the metadata within the QUIC connection ID can be 
encrypted and authenticated. This ensures that untrusted nodes cannot tamper 
with or monitor the metadata. Moreover, leveraging existing encryption 
mechanisms or developing stronger ones, such as deriving a separate key for 
Connection ID protection akin to QUIC header protection, can further improves 
its security.

Despite my effort to mitigate those fundamental concerns, I recognize that 
additional perspectives and insights are invaluable. Therefore, I eagerly 
invite your comments and suggestions on this proposal. I am currently in 
Brisbane this week and would greatly appreciate the opportunity to discuss this 
further and hear your opinions.

Thanks,
Hang

[1] https://datatracker.ietf.org/doc/draft-shi-quic-structured-connection-id/ 
[2] 
https://datatracker.ietf.org/doc/draft-kaippallimalil-tsvwg-media-hdr-wireless/ 
[3] https://datatracker.ietf.org/doc/html/draft-ietf-quic-load-balancers 
[4] https://datatracker.ietf.org/doc/draft-zmlk-quic-te/ 

-----Original Message-----
From: [email protected] <[email protected]> 
Sent: 2024年3月4日 20:05
To: Shihang(Vincent) <[email protected]>
Subject: New Version Notification for 
draft-shi-quic-structured-connection-id-02.txt

A new version of Internet-Draft draft-shi-quic-structured-connection-id-02.txt
has been successfully submitted by Hang Shi and posted to the IETF repository.

Name:     draft-shi-quic-structured-connection-id
Revision: 02
Title:    Structured Connection ID Carrying Metadata
Date:     2024-03-04
Group:    Individual Submission
Pages:    7
URL:      
https://www.ietf.org/archive/id/draft-shi-quic-structured-connection-id-02.txt
Status:   
https://datatracker.ietf.org/doc/draft-shi-quic-structured-connection-id/
HTML:     
https://www.ietf.org/archive/id/draft-shi-quic-structured-connection-id-02.html
HTMLized: 
https://datatracker.ietf.org/doc/html/draft-shi-quic-structured-connection-id
Diff:     
https://author-tools.ietf.org/iddiff?url2=draft-shi-quic-structured-connection-id-02

Abstract:

   This document describes a mechanism to carry the metadata in the QUIC
   connection ID so that the intermediary can perform optimization.



The IETF Secretariat


Reply via email to