- From: Dominique Hazael-Massieux <dom@w3.org>
- Date: Wed, 31 Jul 2019 18:23:57 +0200
- To: public-networks-ig@w3.org
Hi,
Here are the minutes of our Web & Networks IG teleconference held today:
https://www.w3.org/2019/07/31-web-networks-minutes.html
They link to the 3 slidesets that were presented during the call:
https://www.w3.org/2019/07/31-web-networks-PrinciplesForWebNetworks.pdf
https://www.w3.org/2019/07/31-web-networks-agenda.pdf
https://www.w3.org/2019/07/31-web-networks-TheviewfromtheEdge.pdf
See also the use cases in our github repository:
https://github.com/w3c/web-networks/issues
The minutes are copied as text below for reference.
Dom
Web & Networks IG teleconference
31 Jul 2019
[2]Agenda
[2]
https://lists.w3.org/Archives/Public/public-networks-ig/2019Jul/0002.html
Attendees
Present
angel, ChrisNeedham, DanDruta, EricSiow,
JonasSvennebring, JonathanDing, Larry.Zhao, Song,
Songfeng, Sudeep, XiaoweiJiang, YajunChen, DanDruta,
Dom, JetYu
Regrets
Chair
Song, Sudeep, DanD
Scribe
Dom
Contents
* [3]Topics
1. [4]Welcome
2. [5]Principles for Web & Networks solutions
3. [6]Use cases & requirements
4. [7]The view from the Edge
* [8]Summary of Action Items
* [9]Summary of Resolutions
__________________________________________________________
<scribe> Scribe: Dom
Welcome
Sudeep: welcome to the 2nd IG teleconference - we had our first
call a month ago
... TPAC is coming soon
... summer is holiday season in part of the world, so thank you
all for joining
... We have quite a few new participants, so let's quickly go
around the table
Dom: Staff contact for W3C, helped set up the group and
organize Web5G workshop
Song: co-chair for this group, from China Mobile
SongFeng: I'm the AC Rep for 360 tech
Angel: from Alibaba, co-chair of Chinese Web IG, curious to
discover this IG
Song: Jet Yu is from Tencent - first participation to this call
ChrisNeedham: from BBC, first time on this call
... we are interested in the general area of Web & Networks to
understand what this group is doing on how to work with it
... also representing the Media & Entertainment IG - the two
groups should work together and want to help with this
Eric: from Intel, AC Rep and on W3C Advisory Board
... real pleasure seeing this group coming together starting
from Web5G workshop last year
... looking forward to working with all of you
DanD: Dan Druta with AT&T, co-chairing this IG, quite excited
to see how we can progress on the topics
Jonas: from Intel, working with wireless @@@ - interested in
seeing how getting apps and networks to work together in
solving issues
Jonathan: from Intel, based in China working on Web of Things,
and WebRTC
Xiaowei.Jiang: from Xiaomi, happy to join this Web & Networks
IG
... involved in 3GPP 5G, WoT, Web Games, Web apps
... glad to see how the Web can go together with the 5G
network, what use cases AR/VR, Cloud gaming
Sudeep: from Intel, co-chairing this group with Dan and Song
... specialized in networking and virtualization recently
... The agenda for today has 4 topics - not sure how much we
will be able to cover
... first we will cover principles for Web & network solutions,
to be presented by Dan, which will cover some of the background
on the IG
... we will then cover the use cases that have been contributed
so far
... and define how we derive requirements from them
... Dan will then give an update on Mobile Edge Computing
developed in ETSI - to see how that would work from a W3C
perspective
... Song will give an update on liaisons to GSMA, 3GPP
Principles for Web & Networks solutions
[10]Slides: Principles for Web and Networks solutions
[10]
https://www.w3.org/2019/07/31-web-networks-PrinciplesForWebNetworks.pdf
Dan: [Slides: Principles for Web and Networks Solutions]
... at the Web5G workshop last year in London, hosted by GSMA,
we had quite a few interesting conversations on a number of
interesting topics
... we stepped back and had a few conversations and explored
some of the things that had been done in terms of network API
... this led to a breakout session at last TPAC, which showed
interest
... which led to the creation of this IG to channel these
discussions in a more formal way
... I'm going to share what transpired out of these discussions
... as guiding principles for us, to ensure we're relevant to
the community we're participating in
... these principles are summarized in a few buckets:
... * the first one: privacy & transparency
... we've heard many times some network solutions not being so
transparent e.g. in terms of what middleboxes or intermediaries
are doing
... we want to make sure everything we're doing avoids passive
interception, makes clear there is visibility in what gets
shared with whom, with control in the user's hand
... also minimizing fingerprinting for privacy
... * second bucket: trust
... solutions in the past have relied heavily on trust
relationships, and it's hard to deploy at scale
... you have to have a framework in place to validate the trust
... there is an opportunity to build a mechanism that
discourages parties to cheat
... the idea is to have a trade-off where picking one option or
another always comes with both benefits and inconvenients
... * data integrity: we need to make sure that there is a
mechanism in place to guarantee the integrity of the
information that is exchanged
... * focus on hints
... over the years we talked a lot about embedding elements of
data and attributes into flows, like QoS
... focusing on hints that we should look into - that fulfills
some of the principles that I mentioned earlier
... hints that can be validated for authenticity
... something that is not necessarily always expected - some
networks may be able to provide congestion information, but a
Web app should not rely on the availability of that data
... likewise, an app can pass info to the network but should
not expect that this info will always be taken up by the
network
... These are very high level considerations, principles that
emerged from proposals we looked at
Sudeep: the key message is that when it comes to hints, it
should not be expected these hints would always be available
... when we look at use cases and at various network
parameters, we can use these principles as baseline to follow
Use cases & requirements
[11]Slides: Web & Networks Use-cases and requirements
[11] https://www.w3.org/2019/07/31-web-networks-agenda.pdf
Sudeep: thanks to all the folks who've contributed use cases -
11 so far, spanning across 6 to 7 domains
[12]use cases submitted to Networks IG github repo
[12] https://github.com/w3c/web-networks/issues
Sudeep: please feel free to contribute more such use cases
using the github template, or share feedback through comments
... the template guides you on structuring use cases
... we're also happy to receive contributions in any other
format - the chairs can take care of filing the github issue if
needed
... There was a discussion about use cases discussed in other
groups (e.g. WebRTC, or Multi party gaming)
... we will connect with these groups - Dan and Jonathan can
help with WebRTC connections
... we don't want to duplicate use cases, but look at them from
a network perspective
... likewise for Media & Entertainment - content delivery,
broadcast have elements of latency, network quality
... if there are any upcoming use cases you'd like us to look
at, please let us know
Chris: will do
Sudeep: likewise for Web of Things, Smart Cities - we will
reach out to the relevant chairs
... Use cases from other work, e.g. MEC
... some of them would need to brought in, Song will reach out
to ETSI
... all of this to say we're looking at use cases from beyond
this group's specific expertise
Song: [use case for UHD]
... [Use-case #4 Remote Education Service]
... low latency allows synchronized video streaming in rural
areas
... the Web is an ideal platform for resource restricted
computer environment - WebRTC allows to build the Web app for
this
... [Use-case #5 Remote diagnosis]
... [Use-case #7 - Augmented reality advertising and promoting]
[13]UHD - Augmented Reality advertising and promoting #7
[13] https://github.com/w3c/web-networks/issues/7
Song: AR recognition can benefit from cloud-based support for
low-end devices
[14]UHD - Virtual Reality #8
[14] https://github.com/w3c/web-networks/issues/8
Sudeep: taking AR/VR examples, latency is critical
... Mobile Edge Computing may be a critical component of how to
address that aspect
... this looks promising - there may be scope for a Web app to
distribute its computing to the cloud, the edge or on device
... Dan will tell us a bit more about Mobile Edge Computing
The view from the Edge
[15]Slides: The View from the Edge
[15] https://www.w3.org/2019/07/31-web-networks-TheviewfromtheEdge.pdf
Dan: I will cover more than Multi-access Edge Computing (which
what ETSI MEC stands for)
... I want to cover the overall principles, and what the
industry is doing in the space of Edge computing
... ETSI stands at the foundation, but other things are
happening in this space
... in the basic Client / Server architecture, functionality
has moved back and forth between client and server based on
breakthrough in compute, in networkings
... What Edge brings is a third leg - an intermediary piece of
the application sitting between the client and the server
... where it lives is relative to where we see the edge from an
operational perspective
... it could be under your car seat
... it could be in a central office in your local geography
... it's not a new concept - CDNs sort of accomplish that for
static content
... proxy servers with their transcoding and caching
perspectives are of the same sort
... but the big paradigm shit is making applications
edge-aware, edge-location agnostic
... you architect your application to be able to make use of
the edge, independently of where the edge exactly is
... from a client perspective, the edge is a server, but the
edge itself is likely a client to a server
... it all comes down to how it is designed
... from a deployment perspective, the edge is about
administrative boundaries
... but it is also related to network and data center topology
... the edge piece can be moved closer or further from the
server, or from the client
... all of this becomes more relevant with virtualization which
makes moving and deploying pieces much easier
... when you start having all of these edge nodes (in the
1000's), you can't be deploying technicians to manage - it has
to be zero-touch management
... The use cases for Edge cannot be categorized in a single
dimension
... looking at these 3 dimensions (functional requirements,
control and management, network location placement), use cases
will fit in several dimensions
... Edge computing sits at the intersection of several
technology trends - Edge computing gets deployed coupled with
5G
... network slicing and edge cloud are also accompanying that
trend
... How is Edge relevant to the Web, sitting on the client
side?
... most of the edge work is infrastructure-driven - how to
leverage the closer proximity of computing
... but what should a client do to leverage the edge?
... there are initiatives for MEC APIs - most of them are
really addressing the edge application and the infustrcuture
where it is residing
... but there are also discussions on how do you find an edge
node?
... what happens when a client migrates?
... most of these things are running in a client SDK
... we should start looking at how this could be relevant to
Web apps
... we may be able to see some of these APIs for discovery and
validation
... pushed up on the Web layer if there are benefits to it
... the last 2 slides give an overview of the ETSI MEC
architecture, incl links to an overview presentation of
business cases and technical details of available APIs
... and a Linux Foundation open source project, akraino,
focused on vertical integration, defining blueprints for
deployments, from hardware to applications
... Edge may be relevant very relevant to some of the
discussions we will have
... we need to be very crisp on when and how it may be useful
to us
... in particular, between the client and the edge node
... what APIs may need to be exposed on the UA for MEC support?
Sudeep: thank you Dan for the presentation
... I encourage all of you to look at the links and see what
APIs may be useful in the context of use cases
... and bring that to use cases
... we'll be looking at organizing one more meeting before TPAC
Summary of Action Items
Summary of Resolutions
[End of minutes]
Received on Wednesday, 31 July 2019 16:24:02 UTC