Not that this is a downside.
> 2) Looks like its simply a way to setup pipes... not in fact a way to
> marshal and manage computational resources
Right. It is a bit confusing though that they keep saying that this
could be used to deploy active networks, but it doesn't seem to me
that they have any way to cause arbitrary computation at the overlay
routers. They talk briefly about having different forwarding
algorithms for each overlay:
"The overlay can also determine the forwarding algorithm, such as
shortest-path or policy- or QoS-sensitive."
But they don't seem to have thought through any of the issues involved
in making this more than a "you set the flag to pick which one of the
N algorithms that we saw fit to write and deploy" system.
Even as a way to setup pipes, it doesn't look to me like they have a
story for any of the hard problems, like selecting a virtual
topology.
I guess it would be nice to know the extent of their DARPA/NSF
mandate, i.e. do they play to deploy something or just play with this
thing at ISI. And I guess we need to start thinking about new names
(since I don't see this filling the vision of our X-bone anytime
soon).
Andy
-- //------------------------------------------------------------------------- // Andy Collins -- KC6YEY acollins@cs.washington.edu // Graduate Student, University of Washington // Cleveland's Highway Law: // Highways in the worst need of repair naturally have low traffic counts, // which results in low priority for repair work. // (collection last updated by Don Woods) //-------------------------------------------------------------------------