|
|||||||||||
|
Re: [dtn-interest] Bundling Agents
From: Scott Burleigh <Scott.Burleigh(at)jpl.nasa.gov>
Date: Sun Apr 03 2005 - 20:02:28 EDT
Matt Bradbury wrote:
Yes. > Does just one bundling agent reside at each "endpoint identifier". Or
Going with Keith's concept: every bundle agent is a DTN router, but it would be possible for a DTN router not to be a bundle agent. Each is a set of state machines residing on some computer -- not the computer itself. This is a little different from the way we use the terms in the documents, where I think "bundle agent" is the more general term and "DTN routers" are a subclass of bundle agents that aren't just sources and sinks of bundles but may also forward bundles sourced elsewhere. We need to resolve this one way or the other, along with a lot of other terminology issues, but for now I don't think there are any contexts in which we get into trouble by using the terms interchangeably.
> Does a bundling agent control many "demux identifiers" or just one? Or
Each "demux identifier" is analogous to a port number, a token that maps to some sort of device at which an application can use the services of the bundle agent. A bundle agent multiplexes outbound bundle payloads from the devices corresponding to multiple demux identifiers (notionally, multiple application clients) and demultimplexes inbound bundle payloads to them. > In my head it still seems that the "bundling agent" is nothing more than
Yep, same model.
The only
And that is strictly a matter of implementation. There's nothing in the architecture that requires that an application client make any sort of network connection to the bundle agent at which it is registered, and in fact my implementation doesn't operate this way. Scott dtn-interest mailing list dtn-interest@mailman.dtnrg.org http://mailman.dtnrg.org/mailman/listinfo/dtn-interest Received on Sun Apr 3 20:10:04 2005 This archive was generated by hypermail 2.1.8 : Wed Aug 23 2006 - 13:27:00 EDT |
||||||||||
|
|||||||||||