Craig Birkmaier wrote: >http://www.pbs.org/cringely/pulpit/pulpit20050120.html > > Did you read the entire story Bert? > > If Cringely is right, would you call this a walled > garden IPTV deployment, or an effort to BYPASS the > walled gardens? First of all, this isn't IPTV. Unless I missed something in the article, all he's talking about is web sites where one would download HD movies. The sites will be controlled access, but there's no hint that this is an IPTV scheme. (It might be that these web servers are contained in walled gardens, in various ISP nets, but that's not mandatory.) Secondly, my definition of a walled garden is a controlled access *network*. Almost by definition, any network which supports multiple IP Multicast groups is a walled garden. You might want to read RFCs 2365 and 3171 to see what the thinking is. Digital cable is a very good example not only of a walled garden, but also a good model for how a real world IPTV deployment would look, from the perspective of bandwidth allocation and access. Specifically, in a digital cable, you have only so much bandwidth dedicated to Internet access, and the bulk of bandwidth dedicated to high quality TV streams. In an IPTV network, the network would similarly dedicate a small portion of its total bandwidth to Internet access for its subscribers, and the the bulk would go to a pre-selected set of high quality TV stream multicasts. The only significant difference (from a user's perspective) is that in digital cable, the end user gets the entire fire hose delivered into his home, to parse through as he sees fit. In an IPTV network, the local loop is seriously bandwidth challenged, and therefore the network must do the intelligent parsing at some node outside the home. But fundamentally, there's no big difference. The IPTV network will have a set of local IP multicasts streams from which its subscribers can choose. That's why the hype about IPTV is simply overblown and uninformed. IPv6 and much faster network cores could change this picture in theory, but the problems have to do with more than just speed and address space. The bigger problem is settling time of multicast trees and the amount of state information routers would need to maintain to support huge numbers of global multicasts. So I'm not getting any indication that multicasts will be done in dramatically different ways with IPv6. The talk is still walled gardens. Routers in the Internet do not forward multicasts by default, only when and if specific groups are selected. Bert ---------------------------------------------------------------------- You can UNSUBSCRIBE from the OpenDTV list in two ways: - Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org - By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word unsubscribe in the subject line.