So far....
Our strategy: schedule packets in order of GPS completion-time
Our approach: the virtual time function V(t)
- PGPS [Parekh & Gallagher]
- WFQ [Demers, Keshav, & Shenkar]
Bound: fi,WFQk ? fi,GPSk + Lmax / r
-
- i.e., the (real-time) completion time of any packet in WFQ/ PGPS is delayed by no more than the time to service a single packet
So, everything’s just fine....?