Tuesday, September 2, 2008

HE DESIGN PHILOSOPHY OF THE DARPA PROTOCOLS

This was read with great speed before class, as bnrg has been down.

Packet switching because it was "obvious" for the apps, not any high minded idealism
This is strange, as the packet switching model gained them so much
interesting point about priorities. GSM is a network that was designed by industry, so accountability was priority one, which has hampered the network severely.
So there are networks which break the IP plan by providing higher level semantics at the network level. I wonder if that has happened since IP was developed?
good point about IP's failure if network retransmits are common. However, that can also be implemented at a higher level.
Poor implementation of TCP problem was shown with the Savage paper on the fast TCP
Ends with a discussion of TCP. Quite good, and shows how scalability was priority one. As processing power scales faster than population, we may be able to move away from that model...

No comments: