Slashdot Mirror
Is the MBone / Multicast Dead?
"Zow" asks: "I've been looking into a P2P web-caching scheme and noticed that Squid already has the underpinnings for this using multicast, which seems like a reasonable way to do P2P requests. Reasonable, except for the fact that all my web searches for multicast in general and the MBone in particular mostly turn up sites more than 3 years old. Even the MBone FAQ was pulled because it was so old. MBone.com now belongs to a domain squatter. So, what happened? Did everyone give up on Multicast for all practical purposes? Is everyone who was interested in multicast now working on Internet2? Is it only being used for LAN applications? What caused this loss of interest? Cheap bandwidth? Lack of applications? Lack of network support? Unforeseen technical difficulties? Is it still a viable technology for anything?"
Multicast is a delightful concept. Consider that it would make it possible for everyone in the country to watch a streaming video or listen to a music station with one and only one original stream from the source.
The traffic would be carried only by those routers who had downstream listeners, and not carried where there were none. What a delight.
What's the problem? It would require the owners of all those routers to cooperate with each other. I think that's enough to kill it right there. (Yes, there are some technical issues as well)
A.
...bringing you cynical quips since 1998
MBone is basically an IPv4 testbed for applications which will become available to endusers with IPv6. If you're looking for multicast, look for IPv6. Multicast also does not fit in today's business models which tend to see internet providers as downstream for centralized content producers. Often even a fixed IP address is not available with consumer flat rate plans. That should give you an idea where most customers fit into ISPs' plans.
Modern multicast runs natively across and between isps that support it using PIM, MSDP, and MBGP. If anything the portions of the internet wich support interdomain ulticast are far more extensive (and more importantly robust) than they were in the tunneled era...
a list of isp's the support multicast services for customers is at:
http://www.multicast-isp-list.com/index2002.html
It was Slashdotted and the editors keep posting duplicates of the same story.
It is said IP6 has much better multicast support, but I'm not holding my breath to see IP6 either. From my limited perspective, there is ZERO incentive for any ISP to go IP6.
Someone with serious broadcasting interests needs to start pushing the buttons of major ISP's to get them routing multicast to everyone. NBC? ABC? FOX? ClearChannel? Anyone?
I've been ripping TV to mpeg for a while now, and downloading the stuff I miss from various P2P networks, and it's just awesome. Watch what I want, when I want to watch it. This is the future. But all existing P2P networks are unreliable (but props to ed2k for being the least unreliable). Who wants to start a company who will negotiate with content providers for broadcast rights, and transmit it over the internet. Don't do all the DRM crap, but instead let people pay for it, or place a few-minute video ad at the beginning of the content they download. I would do this in a second over P2P networks if it is more reliable (i.e. multicasted).
-- Bob
1^2=1; (-1)^2=1; 1^2=(-1)^2; 1=-1; 1=0.
~~~
posted anonymously because it really isn't all that funny
I've worked on a Mid-2001 online concert featuring the return of a very famous Italian, singer, Mina. The event was available in multicast for people connected to Italian ISPs Inwind and Libero/IOL/Infostrada (totalling around 7 - 8 million subscribers), Unicast for others. I personally configured the front-end routers with the PIM Rendez-vous point :-)
ISPs have merged, pages have changed, but by chance the concert page is still up, although it's in Italian.
Now, I'm working @ a Satellite ISP and we're streaming Multicast all the time (both windows media streams and a data-push protocol which needs no return channel (carousel + heavy FEC)).
Multicast dead?
*not quite*
I'm also about to multicast-stream TV off 802.11b at my house just to see how it works!
P.S.
By the way, I'd HAPPILY kick the last Microsoft boxes out of my server room. Does anybody know of a multicast acquisition-streaming platform which could be a drop-in replacement for Windows Media Encoders + Windows Media Server?
Must work with Windows Media Player on the other side!
Vacuum cleaners suck. Kings rule.
is a simple and quick way to automatically configure multicast systems. Joining a multicast group seems rather difficult ATM, given the fact that you have to get your ISP to set you up. If it was dead simple, something like IPv6 autoconfiguration that happens dynamically on the fly it might be more popular.
just why did you feel compelled to tell us that - I mean, wouldn't we've just -know- who she is, and where's she's from, etc.?
Since then, the term "MBone" came to refer to the subset of the public Internet that is connected via IP multicast routing. (Nowadays, the multicast routing protocol that's most commonly used is PIM-SM.) So, the "MBone" still exists, although it's still not nearly as widespread as many of us would like.
At Carnegie Mellon University, you can request to have your resnet network segment placed on the MBONE. The network folks are pretty good about it -- I asked and was placed on it for a semester. Found that (at the time, three years ago) Linux multicast code was pretty buggy, hard freezing Linux occasionally.
I was interested in writing some multicast stuff, but never got around to it.
Most apps using the MBONE are quite old -- MPEG1 streaming and the like. They also use quite a bit of bandwidth.
The reason no one provides MBONE access is that (a) most of the apps that could use it *eat* bandwidth like crazy and (b) no one asks for it. RoadRunner is not going to put out MBONE access when about three users on their network care.
You might have better luck with a techie-friendly ISP, but maybe not even there...
Oh, and most of the apps that use the thing are pretty old...tk/tcl things that you can manage to get working with some work.
There are some newer protocols (I believe Cisco or Intel has some newer protocols or apps or something like that) that run on the MBONE, but no Linux implementations.
Frankly, last I looked, there weren't a lot of people trying to make it easy to play around with MBONE or the 6bone.
The biggest potential modern uses of the MBONE is probably:
* P2P -- have a server that queues up requests for a file for a day or two or three, and then starts sending all of them at once
* A backbone for video over IP used to replace the current TV networks. Or maybe Internet radio over networks -- if Netscape shipped with a WinAmp with Digitally Imported bookmarked, there might be enough simultaneous listeners for it to be worthwhile for ISPs to use MBONE support.
MBONE is cool in that it's a great equalizer -- a single person can share as much as a massive company can traditionally, with far fewer bandwidth costs. It reminds me of a bit of what the Internet did for traditional media...
May we never see th
there's your answer.
I'd be interested in hearing the results of your research. It seems to me the web is slowly getting bogged down as more and more content centralizes. I'd be more than interested in knowing of and/or contributing to any ongoing projects to implement distributed web servents.
Please consider making an automatic monthly recurring donation to the EFF