RE: [kNT] Multicasting [was -- akamai]

[Rupert Baines <rupert@xxxxxxxxxxxxxxxx>]

 Just to throw my 0.002c-worth in...

Multicast is sort-of real & sort-of available.

It isn't so much a technical issue as a an application.financial one (as
several people have suggested)

Scott said: 
>I believe that some of the more imaginative operators are implementing
multicast and mcast peering but I suspect it will be a long while before
it gets really close to the 'last mile' router.

Actually, most broadband networks deployed today do support it, even in some
cases at the 'last mile router' - the last router (as such) typically being
a BBRAS in a city-POP, which will support muticast; with dumber devices at
the actual last mile (DSLAM, CMTS) which may do. 


Financially...   simplistically, it suits everyone.  The consumer pays $X
for content; N consumers = N * X revenue. If each of those pieces of content
is unicasted, then the bandwisth that the service provider or carrier has to
pay for is proportional to N; if it is multicast they only pay for that BW
once - and then the margin is much higher. Go figure;)  Everyone is happy:
consumer gets what they want (probably with better quality as less traffic =
less contention), service provider makes money. 

But...  this only works if everyone wants exactly the same content at the
same time or you can control enough that it can be cached, which complicates
things & only works if it is static content. 

This can be perfect. For example, Gordon raised the point about satellites,
and this is done, using DVB "containers". For example, Volkswagon send
catalog & spares data every night to factories & dealers in - a few MB of
data they can send over a satellite link instead of posting CD-ROMs.
(strictly it is broadcast, but as outsiders can't decrypt the content, the
effect is the same as multicast)

It would probably work great for BBS.COM streaming the archers, or NBC.COM
streaming the olympics
But I'm not sure there are enough of them yet to make it worth implementing
- unless there's a clear revenue model.  For example, telcos who want to
deploy VDSL so they can offer "broadcast" TV - which would actually be
multicast.

But if people want different content  (which is the intent of a lot of
personalisation; Amazon explicitly wants me to see a different home page to
the one you get), or nominally the same but at different times (different
news, ads) then caching & smart servers & CDNs are probably the appropriate
choice.