Dear All,
In response to Dave's message, some comments (as he requested) and other
thoughts. It took a while to put together: possibly because there are so
many ways of skinning this particular herd of cats... I did promise them
earlier so sorry for the delay.
I'm trying to focus strategically - what knowledge base do we want to build
- not tactically - how to get it.
Social: we have created a new ICT ecosystem (yuk, but I suppose that's the
best way to describe it) that let's people create multiple communities
(defined by a common purpose, good and bad), interconnected, overlapping,
etc., just as they want to. If I understood some of Dave Clark's recent
thoughts correctly, we should be concerned about what is decent, honest and
truthful, legal and ethical. This may relate to what Jon said about good
guys and bad guys. There is a massive problem of coexistence and the
solution will come from the people involved, hoping they don't start
fighting wars - even if that's one outcome.
Technical: virtualisation is back. We have it at layers -1 to 7,
(wavelengths, all those things they do with OFDM/CDMA etc. on wireless,
provider VLANS, closed user groups, user VPNs, the GRID, agent systems,
etc. etc.) and in end hosts (hardware, firmware as well as software). It's
a way to support and constrain those communities, but how?
Economic: I would put multicast, QoS and security on the spot here. Also
transport protocols. They cost a lot, they can be done or not done and you
pay a price either as operator or user. They can be done at any layer. What
combination works for a given purpose? Some resources may be accessible to
all but they aren't free (except when you don't pay). How does this work
financially?
Mechanisms: naming (of objects or their type), addressing (where they are
right now), discovery and routing are what make the ecosystems work. Do we
want one architecture or many? Transport: DCCP, UDPLite, SCTP, and all the
rest. This is all research bread-and-butter, maybe not strategic though
when building a programme.
What not to do: cross layer optimisation.
I hope this helps, and I did put my hand up...
Good luck to Dave in his discussion with Steve.
Regards,
Alistair
--
Prof. Alistair Munro, Toshiba Professor of Communications Networks
Dept. Electrical & Electronic Engineering, Bristol University
Rm 4.29, Merchant Venturers Building, Woodland Road, Clifton,
Bristol BS8 1UB
E-mail: [log in to unmask]
Office Tel: +44-117-954-5655; Fax: +44-117-954-5206; Cell/Mobile:
+44-7974-922442
|