Hi Shane,
> for 'live' content, given the inherent efficiency of multicast.
I am not questioning the efficiency of multicast. In fact I am looking
at peercasting and can't find it's inefficiencies either provided it is
deployed in a proper manner.
I am not talking about residential hosts replicating the live content.
But if you place your peercasting servers in the key network locations I
could argue that peercasting can be as efficient as multicast both in
core as well as distribution levels.
The difference between the two is just the way such distribution tree is
constructed. In multicast you have protocol which in most cases (not
talking about p2mp te) follows the unicast routing to reach the src.
On the other hand in peercasting you can use many other elements like
RTT, feedback from ALTO servers, routing proximity, sources load etc ...
for the tree building decision which may be hard to do by multicast.
While you are right that for live streaming multicast may be the winner
there is IMHO much bigger market for on demand file, video,audio
streaming where caching may be of huge importance. And I guess this is
very clear that hosts/servers are much better in content caching then
routers.
Cheers,
R.
Robert,
On Nov 29, 2009, at 03:48 MST, Robert Raszuk wrote:
Tony,
Isn't it a case that multicasting has been already obsoleted in the
quite significant degree by peercasting ?
http://www.viswiki.com/en/Peercasting
IMHO, I think it's rather difficult to predict that peercasting
has/will obsolete multicast, at least in the U.S. and, specifically,
for truly "live" content, e.g.: sports, news broadcasts, etc.
Unfortunately, the usual chicken-and-egg problem exists: no widely
appealing, legitimate content exists, therefore there is no demand,
hence very few SP's set-up their infrastructure to provide multicast
down to CE devices. However, with the dearth of *unicast*
time/place-shifted video content coming online by the major players
(cf: Hulu, Apple, Netflix, Blockbuster, etc.), it may be a matter of
time, before we start to see a growing demand for *multicast* video
for 'live' content, given the inherent efficiency of multicast.
Furthermore, while peercasting is an intriguing application-level
form of multicast, (therefore, allowing a content owner/distributor
more control over when & how they distribute their content w/out
having to wait for the network to be ready), it's not without it's
"challenges", specifically: - asymmetric BW on residential access
circuits, making it difficult to hairpin/push content back up into
the network; - lack of some form of CoS on lower bit-rate residential
lines to either: a) assist with pushing hair-pinned content back up
into the network; and/or, b) avoid interfering with other legitimate
apps on the access circuit, (e.g.: VoIP, etc.) - etc.
I guess time will tell.
-shane
And I am not that convinced that replication at network
router/switch level is that much more efficient then replication at
the network host level. I think it is just a matter of efficiently
determining the optimal replication points.
Cheers, R.
Dino Farinacci wrote:
Well if multicast proper is not on the requirements for scaling
the routing architecture, how can we set our sights higher?
Do you expect multicast route scaling to be solved in some
other forum like perhaps SAMRG?
I think you can just work on it. No permission is required. But
first you should take this up with the guys who have been working
on multicast for the last 17 years. ;-) Seriously tho, if that's
a useful work item, the RG can take it up after this work item
is complete. Tony _______________________________________________
rrg mailing list [email protected]
http://www.irtf.org/mailman/listinfo/rrg
_______________________________________________ rrg mailing list
[email protected] http://www.irtf.org/mailman/listinfo/rrg
_______________________________________________
rrg mailing list
[email protected]
http://www.irtf.org/mailman/listinfo/rrg
