Re: [dtn-interest] Custody and Fragmentation

From: Scott Burleigh <Scott.Burleigh(at)jpl.nasa.gov>
Date: Wed Apr 13 2005 - 06:04:20 EDT

Stephen Farrell wrote:
>
> We've beaten ourselves up about reactive fragmentation & crypto
> for the last day or two.

In most cases there's nothing special to say: you transfer custody of a bundle, and a fragment is just a bundle. All the normal custody transfer rules apply.

But there is one additional wrinkle. Suppose the current custodian of a bundle is A and the next router is B, which fragments the bundle but does not take custody. The next router, C, receives one of the fragments and decides to take custody of it; C sends a custody acceptance signal for this fragment to the current custodian, which is A [fragmentation didn't alter custodial responsibility]. Agent A receives a custody acceptance signal for a bundle that is a fragment and is therefore different from the one for which it is the custodian; A must do its own "responsive" fragmentation of the bundle on the boundaries defined by the fragment, so that it can relinquish custody of the fragment but retain custody of the remainder of the original bundle.

Scott

---

dtn-interest mailing list
dtn-interest@mailman.dtnrg.org
http://mailman.dtnrg.org/mailman/listinfo/dtn-interest Received on Wed Apr 13 06:10:42 2005