On 10/11/20 13:58, Rob Janssen via 44Net wrote:
I agree that would be much more efficient than having the basic star-type structure where everyone gets the same stream from one server. That would be one thing that could be implemented in AMPRnet to show the world that it really has advantages.
Exactly my thinking.
Hams (and open source people in general) have a secret weapon: we're willing to work on projects just because we find them technically interesting. We don't have to first convince senior management that they can soon make a lot of money from it. Better yet, we don't have to care if we destroy somebody's existing business model.
Case in point: the Internet. It's actually kind of amazing that I was allowed to spend so much time in the IETF and promoting IP back at Bellcore in the 1980s given that I was really working to destroy my own company. I got strong support from my immediate management, but we were probably lucky in failing to get senior telephone company execs to understand the Internet. When they finally did, in the late 1990s, it was too late. Bellcore's remains are owned by Ericsson (I think) and the building in New Jersey where I worked in the 1980s was sold long ago. The situation was much the same with Xerox PARC (see "Fumbling the Future").
A really good, scalable wide area IP multicast architecture has the potential to destroy a lot of existing business models, including content distribution networks (CDNs), many commercial conferencing services (like Zoom) and maybe even some mega application-level services like Facebook, Twitter and Youtube. They've become so huge and successful in part because there are strong economies of scale in the resources needed to throw terabits of fiber and buildings stuffed with computers at the task of centrally replicating and transmitting unicast packets. But if those packets no longer have to be centrally replicated...
For phone quality voice and the number of attendees we would expect, running this over AMPRnet in the star configuration should be no issue.
Well....yes. But that's not the point. We need to show that wide-area multicasting can be done, and that'd be easiest if you start with a light load like voice.