Re: Proposal for Integrating Deadline-Aware Streams into Multipath QUIC using DMTP

Tony John <tony.john@ovgu.de> Thu, 07 November 2024 08:13 UTC

Return-Path: <tony.john@ovgu.de>
X-Original-To: quic@ietfa.amsl.com
Delivered-To: quic@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D509FC151535 for <quic@ietfa.amsl.com>; Thu, 7 Nov 2024 00:13:05 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.895
X-Spam-Level:
X-Spam-Status: No, score=-1.895 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_BLOCKED=0.001, RCVD_IN_ZEN_BLOCKED_OPENDNS=0.001, SPF_HELO_NONE=0.001, SPF_PASS=-0.001, T_KAM_HTML_FONT_INVALID=0.01, T_SCC_BODY_TEXT_LINE=-0.01, URIBL_DBL_BLOCKED_OPENDNS=0.001, URIBL_ZEN_BLOCKED_OPENDNS=0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([50.223.129.194]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id xFBVgLEEoMze for <quic@ietfa.amsl.com>; Thu, 7 Nov 2024 00:13:01 -0800 (PST)
Received: from mail.ovgu.de (mail.ovgu.de [141.44.1.66]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (P-256) server-digest SHA256) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 3F272C151080 for <quic@ietf.org>; Thu, 7 Nov 2024 00:12:59 -0800 (PST)
Received: from [10.1.0.7] (unknown [141.44.25.153]) (using TLSv1.3 with cipher TLS_AES_128_GCM_SHA256 (128/128 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by mail.ovgu.de (Postfix) with ESMTPSA id 232D4A0A3C; Thu, 7 Nov 2024 09:12:57 +0100 (CET)
Content-Type: multipart/alternative; boundary="------------nqWevXdUAK6zB8nuDNhpsNZd"
Message-ID: <68edcfba-11f5-4393-a2e5-942ef48a7d45@ovgu.de>
Date: Thu, 07 Nov 2024 09:12:55 +0100
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: Proposal for Integrating Deadline-Aware Streams into Multipath QUIC using DMTP
To: Ian Swett <ianswett@google.com>
References: <01f5c2ad-2129-495d-9385-5bcdc65d7467@ovgu.de> <CAOYVs2pDMpgWtUb48mL8phAJbrkZ826xGf94nO-BgPjLfnEhzQ@mail.gmail.com> <CAKcm_gMkgaBp6NeNBEEMqBTLQuMgHAnH=Q6x8xHRs-2dqiBFEQ@mail.gmail.com>
Content-Language: en-US
From: Tony John <tony.john@ovgu.de>
In-Reply-To: <CAKcm_gMkgaBp6NeNBEEMqBTLQuMgHAnH=Q6x8xHRs-2dqiBFEQ@mail.gmail.com>
Message-ID-Hash: WAF73HA4O2RWDSKPO4OZPCUU2UQSH37H
X-Message-ID-Hash: WAF73HA4O2RWDSKPO4OZPCUU2UQSH37H
X-MailFrom: tony.john@ovgu.de
X-Mailman-Rule-Misses: dmarc-mitigation; no-senders; approved; emergency; loop; banned-address; member-moderation; header-match-quic.ietf.org-0; nonmember-moderation; administrivia; implicit-dest; max-recipients; max-size; news-moderation; no-subject; digests; suspicious-header
CC: quic@ietf.org, David Hausheer <david.hausheer@gmail.com>
X-Mailman-Version: 3.3.9rc6
Precedence: list
List-Id: Main mailing list of the IETF QUIC working group <quic.ietf.org>
Archived-At: <https://mailarchive.ietf.org/arch/msg/quic/mU1egYNrT5-jAp_zfliCoKt_UfM>
List-Archive: <https://mailarchive.ietf.org/arch/browse/quic>
List-Help: <mailto:quic-request@ietf.org?subject=help>
List-Owner: <mailto:quic-owner@ietf.org>
List-Post: <mailto:quic@ietf.org>
List-Subscribe: <mailto:quic-join@ietf.org>
List-Unsubscribe: <mailto:quic-leave@ietf.org>

Hi Ian,

Thank you for your feedback. While it's possible to implement 
deadline-aware streams in Multipath QUIC using implementation-specific 
APIs, this approach may lack endpoint coordination because deadlines are 
not communicated between endpoints through the protocol. By introducing 
a transport parameter and a custom frame, endpoints can negotiate 
support and exchange deadline information directly within the protocol, 
enabling coordinated scheduling decisions at the transport layer. 
Standardizing this mechanism avoids the limitations of 
implementation-specific solutions, promoting wider adoption and 
interoperability of deadline-aware streams across different 
implementations of Multipath QUIC.

Best regards,

Tony

On 10/31/2024 14:23, Ian Swett wrote:
> I'm unclear what is needed from QUIC or Multipath QUIC?  One can 
> already implement deadline aware streams and expose that as an API, 
> but I don't see why the wire format would need to change or why a 
> transport param would be necessary?
>
> Doing deadline aware scheduling over multiple paths seems like a 
> difficult problem to me, and likely better suited to research than the 
> IETF, but I'm not an expert.
>
> Ian
>
> On Thu, Oct 31, 2024 at 7:50 AM Marten Seemann 
> <martenseemann@gmail.com> wrote:
>
>     The paper is paywalled. I don't have access, and I'd assume that
>     this applies to many participants on this mailing list.
>
>     On Thu, 31 Oct 2024 at 17:12, Tony John <tony.john@ovgu.de> wrote:
>
>         Dear QUIC Working Group,
>
>         I would like to initiate a discussion on integrating
>         deadline-aware streams as an extension to the Multipath QUIC
>         protocol. Given the increasing demand for real-time
>         applications such as teleoperation, live video streaming, and
>         online gaming, there's a growing need for transport protocols
>         that can efficiently handle strict latency requirements.
>
>         Motivation
>
>         Multipath QUIC enhances performance by utilizing multiple
>         paths simultaneously, but it currently lacks mechanisms to
>         guarantee data delivery within specific timeframes.
>         Introducing deadline-aware streams to Multipath QUIC could
>         enable applications to meet stringent latency constraints,
>         optimizing for low-latency and high-reliability scenarios.
>
>         Additionally, the ability to have multiple paths using the
>         same 4-tuple opens up the possibility of leveraging paths from
>         path-aware networks like SCION, source routing, and others.
>         This expands the pool of available paths beyond traditional
>         IPv4 and IPv6 routes, potentially increasing the effectiveness
>         of deadline-aware mechanisms like those proposed in the
>         Deadline-aware Multipath Transport Protocol (DMTP).
>
>         Relevant Discussions
>
>         I would like to acknowledge the previous discussion
>         <https://mailarchive.ietf.org/arch/msg/quic/Ie6Ju6cNCocNNktldJJrvZfsXQ8/>on
>         adding deadline awareness to QUIC. The discussion indicates
>         interest in deadline-aware mechanisms and their applicability
>         to QUIC.
>
>         Proposal
>
>         Building upon these ideas, I propose integrating
>         deadline-aware streams into Multipath QUIC as an optional
>         extension. The key aspects of the proposal are:
>
>          *
>
>             New Transport Parameter and Frame Type: Introduce a new
>             transport parameter to signal support for deadline-aware
>             streams during the QUIC handshake and define a new frame
>             type called DEADLINE_CONTROLto signal deadlines for
>             specific streams.
>
>          *
>
>             Leveraging DMTP Concepts: Utilize strategies from the
>             Deadline-aware Multipath Transport Protocol (DMTP), such
>             as smart retransmissions and Forward Error Correction
>             (FEC), to optimize packet delivery based on latency deadlines.
>
>          *
>
>             Custom Scheduler and Congestion Controller: Implement
>             DMTP's mechanisms as a custom scheduler and congestion
>             controller within the Multipath QUIC framework.
>
>         How DMTP Fits In
>
>         DMTP is tailored for deadline-sensitive communication over
>         multiple paths. Its key concepts include:
>
>          *
>
>             Path Optimization: Dynamically selecting paths based on
>             metrics like latency, bandwidth, and packet loss,
>             complementing Multipath QUIC's ability to manage multiple
>             paths effectively.
>
>          *
>
>             Adaptive FEC: Integrating FEC to reduce the need for
>             retransmissions, enhancing Multipath QUIC's congestion
>             control mechanisms.
>
>          *
>
>             Smart Retransmissions: Retransmitting packets only if they
>             are predicted to meet the deadline, avoiding unnecessary
>             retransmissions and improving efficiency.
>
>         For more detailed information on DMTP, please refer to our
>         paper
>         <https://doi.org/10.23919/IFIPNetworking57963.2023.10186417>.
>         We have developed a prototype implementation of DMTP, which we
>         plan to open-source shortly.
>
>         Request for Feedback
>
>         I am interested in the community's perspective on this proposal:
>
>          *
>
>             Value of Exploration: Do you see value in exploring
>             deadline-aware streams as an extension to Multipath QUIC?
>
>          *
>
>             Potential Challenges: Are there potential challenges or
>             compatibility concerns we should be aware of?
>
>         I would greatly appreciate any thoughts or guidance on how
>         best to proceed. Thank you for considering this proposal. I
>         look forward to your feedback and the possibility of
>         discussing this further.
>
>         Best regards,
>
>         M.Sc. Tony John
>
>         Research Associate
>
>         Otto-von-Guericke-University Magdeburg, Germany
>