[MBONED] Multicast to the Browser

["Holland, Jake" <[e-mail]>]

Hi mboned,

I have finally posted a few docs that give an outline of the architecture
we're proposing to make multicast safe enough to run arbitrary interdomain
traffic, and to use it on web pages.

This is a slightly more concrete version of the hand-waving I did at the
meeting in Montreal, and hopefully explains what we're aiming for a little
more clearly.


WICG proposal:
https://discourse.wicg.io/t/proposal-multicastreceiver-api/3939


That refers to a few IETF drafts.  I'd like to request some time in Singapore
to go over them.

I'm also requesting feedback about these docs at your earliest convenience,
as well as any comments you've got on whether these I-Ds are appropriate
for adoption in mboned (and if not, what you'd suggest instead):

1. DORMS (or: how to get metadata about multicast groups from senders):
https://tools.ietf.org/html/draft-jholland-mboned-dorms-01

I'll argue that this is in-scope for mboned, and hopefully not too hard
a read.  Its main point is to provide a vehicle for the metadata for the
other 2 docs, plus other future extensions.


2. CBACC (or: how to ensure the subscribed capacity level stays safe):
https://tools.ietf.org/html/draft-jholland-mboned-cbacc-00

Although I think this doc will need review from TSV, I believe early work
and perhaps adoption is appropriate within mboned, since it's targeted at
enabling interdomain multicast support.


3. AMBI (or: how to authenticate the traffic, assuming you can send
some out-of-band hashes securely):
https://tools.ietf.org/html/draft-jholland-mboned-ambi-04

This one will of course need security review, and perhaps would be best
done in the security area, especially if we can re-open msec.

However, when we last presented the pre-split AMBI+ALTA proposal in
secdispatch, they said we should go find more people who are interested
and come back.

In my ideal world, I'd like to go back to secdispatch, ideally in Singapore,
with some consensus from mboned that this is a worthwhile direction that
would be valuable if it can be made to work (or to get advice from mboned
that it's not, if it's not).

So to anyone who can read the doc and post an opinion on that point, I'd be
very grateful.


Cheers,
Jake