Again, please don't pass this around outside UW.
The Ion Stoica paper looks very cool (although quite dry). Its an extension
of his core-stateless fair queuing work that implements service guarantees
using only edge-based classification and some extra state put into each
packet. Basically a distributed version of virtual clock (the main trick is
figuring out how to avoid precomputing all the deadlines along the path).
If I had to vote for a best student paper, I'd guess its this one (although
I haven't read all the papers yet ;-).
The load-sensitive routing paper is relevant to detour. Its about how to
design stable algorithms that adaptively route long-lived ip traffic (they
concentrate on long-lived flows because they don't want to deal with the
variability of short flows... personally, I wonder if this matters alot).
- Stefan