- From: Dominique Hazael-Massieux <dom@w3.org>
- Date: Mon, 18 Sep 2023 16:49:52 +0200
- To: "'public-networks-ig@w3.org'" <public-networks-ig@w3.org>
Hi,
The minutes of our hybrid meeting at TPAC held on September 11 are
available at:
https://www.w3.org/2023/09/11-web-networks-minutes.html
and copied as text below.
Dom
Web & Networks Interest Group
11 September 2023
[2]Agenda. [3]IRC log.
[2] https://www.w3.org/wiki/Networks/TPAC2023
[3] https://www.w3.org/2023/09/11-web-networks-irc
Attendees
Present
@@@_WoT, DanDruta, DapengLiu, DomHazael-Massieux,
EricSiow, GeoffGustafson, Geun-Hyung_Kim, HuaqiShan,
LinLi, LouayBassbouss, MichaelMcCool, Mirja_Kühlewind,
PiersOHanlon, ShiLing, SongXU, SteveHarrison, Sudeep
Regrets
-
Chair
Dan, Song, Sudeep
Scribe
dontcallmeDOM_
Contents
1. [4]Use Cases and Requirements
2. [5]Next steps on Client/Edge/Cloud coordination document
3. [6]Open Discussions
4. [7]Wrap up
Meeting minutes
Dan: [introducing the Web & Networks IG meeting]
Slideset: [8]https://github.com/w3c/web-networks/blob/main/
meetings/W3C_WNIG_TPAC2023_V1.pdf
[8]
https://github.com/w3c/web-networks/blob/main/meetings/W3C_WNIG_TPAC2023_V1.pdf
Slideset: [9]https://lists.w3.org/Archives/Public/www-archive/
2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf
[9]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf
[10][Slide 4]
[10]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=4
Dan: the IG is missioned to explore solutions for Web Apps that
leverage network capabilities
… this included enabling apps to provide hints to the network
on their needs, and vice-versa
[11][Slide 5]
[11]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=5
Dan: more recently, our focus has been on 3 specific areas:
… - edge computing - even though it remains unclear how edge
computing will materialize from a business & operational
perspectives, we feel it's important we get the technical
aspects understood in advance
… - network quality monitoring & prediction
… - network emulation
[12][Slide 6]
[12]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=6
[13][Slide 7]
[13]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=7
[14][Slide 8]
[14]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=8
Dan: for edge computing, we've developed use cases to
understand the intersection between browsers and edge computing
… and have explored different approaches that would allow
offloading computing tasks
… involving web assembly, web workers, etc
… surfacing those in a technical note
[15][Slide 9]
[15]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=9
[16][Slide 10]
[16]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=10
[17][Slide 11]
[17]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=11
[18][Slide 12]
[18]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=12
Dan: we haven't done much lately in that space; there was
interest on this, but we would still need to put a package
together similar to what we did for edge computing
[19][Slide 13]
[19]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=13
[20][Slide 14]
[20]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=14
[21][Slide 15]
[21]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=15
Dan: Edge computing so far has received the biggest focus of
the group
[22][Slide 16]
[22]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=16
Dan: our current charter ends at the end of October, we're in
the process of rechartering - please comment on the proposed
draft charter
[23][Slide 17]
[23]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/W3C_WNIG_TPAC2023_V1.pdf#page=17
Dan: there are interesting intersections with ongoing
discussions e.g. around energy saving and efficiency
Geoff: can you say more about the split browser model?
Dan: way back in the ways when mobile bandwidth was limited,
there was this approach of a split browser where an element in
the network did some pre-rendering before sending it back in an
efficient format to the end client
… this is more of analogy - e.g. for situations where taking
advantage of more GPU or CPU powers from the network
Dom: the TR publication of the IG Note is awaiting a last round
of edits on the conclusion
Use Cases and Requirements
Slideset: [24]https://lists.w3.org/Archives/Public/www-archive/
2023Sep/att-0009/
WNIG_UC_Study_-_Ubiquitous_Computing___________________________
__________________v5.pdf
[24]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf
[25][Slide 1]
[25]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=1
[26][Slide 2]
[26]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=2
[27][Slide 3]
[27]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=3
Song: SUL - Super UpLink
… most usages focus on downlink
… but there are a number of scenarios that rely on uplink
capabilities
… that has been a focus on some discussions in the 5G and 6G
context
… e.G. manufacturing scenarios, or immersive broadcasting
… this is not just about increased bandwidth - it needs
collaboration with the application layer and other actors of
the ecosystem
… 5GA (5G advanced) will introduce cloud edge and improvements
in the client
… in the context of the Web, WebRTC isn't always a proper
solution - e.g. RTMP is still used for uplink streaming
… HLS is used for downlink streaming - what role for
WebTransport?
… looking at the end 2 end picture, biggest delays come from
the RTMP push and buffering of adaptive streaming
Michael: is this related to the use cases in the edge use case
doc?
Song: this was a use case presented last year
Michael: see [28]https://w3c.github.io/
edge-computing-web-exploration/#UC-SA
[28] https://w3c.github.io/edge-computing-web-exploration/#UC-SA
Song: with transports based on TCP, a weak network is going to
have a big impact
… also H265 isn't available with webRTC
[29][Slide 4]
[29]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=4
[30][Slide 5]
[30]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=5
[31][Slide 6]
[31]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=6
[32][Slide 7]
[32]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=7
[33][Slide 8]
[33]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=8
[34][Slide 9]
[34]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=9
[35]WasmEdge
[35] https://wasmedge.org/
[36][Slide 10]
[36]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=10
[37]Low Latency, Low Loss, Scalable Throughput (L4S) Internet
Service: Architecture
[37] https://www.ietf.org/archive/id/draft-ietf-tsvwg-l4s-arch-19.html
[38][Slide 11]
[38]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=11
[39][Slide 12]
[39]
https://lists.w3.org/Archives/Public/www-archive/2023Sep/att-0009/WNIG_UC_Study_-_Ubiquitous_Computing_____________________________________________v5.pdf#page=12
Michael: we should try to consolidate these use cases with our
existing ones
… do these new use cases fit in existing categories or do they
need new ones?
… do they bring new requirements?
… do they require involvement of new stakeholders / new
collaboration requirements?
Dom: the scope is different - not all the described use cases
rely on edge computing technology
… let's only bring them those that are in the intersection
Dan: right, but we should make sure the ones that overlap get
documented in our existing doc
… and the ones that don't, we should look at doing a similar
exercise for that other scope
Michael: maybe we need a bigger picture explainer on
relationship between edge computing and network capabilities
… or a broader use cases document for the group
Sudeep: when we started the IG, we had 3 streams: edge
computing, network link prediction and monitoring, and tools
… now we're at looking at next generation use cases - we may
need to draw a venn diagram with our existing workstreams and
workstreams in other standardization landscapes
Song: +1 to what Dan said in terms of integration and
complementing the use cases
Michael: it would be interesting to identify which new
requirements are raised by these use cases
Dan: an open question - our next steps include figuring which
stakeholders are needed around the table
… particularly key in the open environment of the Web where
pre-existing trust relationships can't work as in more
controlled environments
… understanding how they would work in the context of the
loosely coupled relationships of the eb Platform
Dom: let's make sure we identify clear next steps to follow up
on these ideas
Michael: our current doc is not finished for sure; e.g. the
trust model or negotation of quality assurances
Max: I would like to discuss next steps on cloud coordination
use cases
Dan: Dom was asking what do we want to do next based on what
has been presented
… if there are things to be pushed to the edge explainer,
that's a reasonably clear step
… for the others, who would be spearheading that new document?
… is this something you could be driving Song?
Song: will be happy to look into this with the support of the
chairs
Dom: thanks Song! getting some inspiration on the structure of
edge use cases, not necessarily focusing on pushing them in the
edge use cases doc
Dan: this matches what Michael is suggesting: creating a
different "advanced network use cases and requirements"
documents
… let's make sure we have robust stories for these use cases
… in particular taking into account the loosely coupled
environment of the Web
Michael: a lot of these questions depend on identify
frameworks, incl (but exclusively) for people
… DIDs may be part of the solution
Next steps on Client/Edge/Cloud coordination document
Max: I think the document is pretty stable
… and we should be able to publish it once we fix the last
remaining issue
… I think forming a CG would be a good way to incubate a
solution in this space
… before proposing a WG
… after the publication of the document as an IG note, we can
start to establish a CG
… we need a title, a short description and 5 supporters to
launch the group
… I'm happy to prepare this if there is support
Michael: the advantage of a CG is that it allows to broaden the
input to non W3C Members
… this has proved useful in the WoT
… we will just have to be clear on IP expectations
Eric: +1, esp since W3C no longer has as many telco Members as
it used to
… CG contributions still require RF licensing
… W3C & 3GPP have very different licensing obligations, I'm not
sure how to reconcile between the 2
DanD: you're right that some of the telco expertise moved away
from W3C; that said, edge computing is much less telco oriented
… it really has to do with workload offload - you probably
wants more of the startup, innovators in the space of
distributed computing than the telco space which tends to be a
lot more rigid in the space of distributed computing
… my only concern with CGs is that they can be like
firecrackers - they too often fizzling up; keeping the momentum
going is something that needs to be considered
Michael: you also don't get staff support
Dom: no issue with creating a CG from my perspective, but
getting people to join and actively participate in the CG will
remain the main challenge
… this raises the question of who would be chairing and
animating it
Song: I support the new CG but indeed we need to be thoughtful
about engagement
Dan: indeed, figuring out the stakeholders, the target
community feels critical
Max: thanks - this is all useful and valid input
Dan: in summary, positive feedback on the proposal, but more
details needed before pushing this idea forward which we hope
Max will pursue
Open Discussions
Dan: the future topics slide may be a trigger for more
discussions
Michael: the name of our group doesn't really surface that we
work on the topic of edge computing
… maybe we could integrate it more clearly in the name
Dan: that makes sense
Sudeep: sounds reasonable; not sure how this would relate to
the other discussion on a CG
Dom: this is about surfacing the edge in the IG name, while
keeping the broader scope; the CG would be focused on solutions
based on the edge use cases document
Sudeep: that sounds like a good proposal to me too then
Michael: Quality of Services negotiation hasn't surfaced
… other subtopics under edge include workload packaging
DanD: re network QoS (e.g. via network slices), it has to do
with the trust model
… which is hard to attack in the context of the Web model
Michael: QoE would need to consider both communication time,
computing time
Wrap up
Dan: Action items:
… Song to go to the next level of details on use cases as a
premise to mapping them into existing or new documents
… Everybody to review the IG new charter and get it supported
… Someone to include "Edge" in the name of the group
… Max to propose an outreach strategy for a potential new CG in
the space of client/edge/cloud coordination
Minutes manually created (not a transcript), formatted by
[40]scribe.perl version 221 (Fri Jul 21 14:01:30 2023 UTC).
[40] https://w3c.github.io/scribe2/scribedoc.html
Received on Monday, 18 September 2023 14:49:57 UTC