🚀 go-pugleaf

Dear John,

You're calling our baby ugly.  Please stop :-) Cable & Wireless internet
run a Joint Venture with Real Networks called the Real Broadcast Network.
We source quite a bit of unicast traffic from this network.  By comparison our
network is currently between 5 and 13 by OC-12 per pop.  I don't know how
many millions of customers it ultimately serves.

There is a very good reason to build a service which delivers unicast video
content.  It's called 'on-demand' delivery.  Multicast is a poor medium for
on-demand delivery of content.  The game in on-demand is to move the
content to the edges of your network thereby giving the end user a low-latency,
'quality' service.  It's difficult to talk in terms of quality video when most users
are connected to the network at 56kbps :-) Quality is moving forward as fast as the
internet can peddle and i expect it to be a service that will someday rival current
broadcast mediums.

Let's remember that multicast is a tool, not a holy grail.  We certainly use
multicast to link the unicast delivery sites, long-haul bandwidth is expensive,
intrapop bandwidth is free.  Multicast is a rather important tool in my toolbox.
I value it greatly as such but i'm not about to let the notion of  'ubiquitous
multicast deployment' stand in the way of offering real products.  In 1995,
while we were MCI, I and others built one of the largest multicast networks
in service.  It may still be one of the largest.  It does not run on the same
routers that route unicast traffic, it is an overlay network.  Why?  because one
of the biggest issues that face ISP's today is platform reliability.  We cannot
afford to introduce another service into an already 'unstable' platform (if
you read between the lines here, platform=router).  My
point is that we are driving multicast deployment as fast as we feel we can.

Through the RBN, we also multicast live content for those few users that are
able to receive it.  our experience with this network shows
that 50-70% of traffic is on-demand. That's a rather large segment of the
service that one can't afford to ignore.

Jon Crowcroft wrote:

> In message <361C37ED.5825@cisco.com>, John Meylor typed:
>
>  >>> perfectly honest there is still some question as to MBONEs ability to compete
>  >>> against Real Video and its ilk in a commercial environment.
>
>  >>These are not competing technologies - multimedia applications can and do utilize
>  >>multicast services - and the better quality these multicast services are - the better off the
>  >>quality of the transmission, whatever the application.
> exactly - well said
>
> i would have a hard time seeing how significant amounts of Real Video
> would be allowed to be _unicast_+ to multiple simulataneous receivers
> on any sensibly run ISPs net....certainly we wouldn't allow it on ours
> and we have 155Mbps access per site and >1M hosts....
>
>  >>While it is, for
>  >>> the most part, effective for talks, classes and the like it would have limited
>  >>> usefulness for most internet consumers.
>  >>
>  >>Wrong. Do you want to pass 10,000 copies of the next Starr report or just one?
>  >>How about the daily news? Stock quotes? The John Glenn mission?, your next
>  >>app upgrade? a training class?   Lot's of people are using multicast in
>  >>innovative ways to conserve network resources wisely, this can be done even
>  >>without an external business reason, it's just smart engineering.
>
> great - couldnt say better...
>
>  >> However, providing MBONE broadcast
>  >>> connectivity for corporate presentations or as some sort of value added
>  >>> service might be feasible.
>
>  >>Multicast is being used widely in the enterprise space already.
>
> absoultely - e.g. share dealing...
>
>  cheers
>
>    jon