Paper Review:
Analysis and Simulation of a Fair Queuing Algorithm
Reviewer: Mark Meras (mm446)
Main Contribution
This paper discusses a fair queuing algorithm designed to be used on
network gateways. Most of the gateways uses First-Come-First-Serve
queuing to determine bandwidth, promptness, and buffer space. The paper
introduces a fair algorithm based on a round-robin scheme that tries to
satisfy the most users, and presents simulation data
Key Points
- Congestion control can happen at the source and at gateways. In
gateways, congestion control is implemented through routing and queuing.
While queueing algorithms do not decrease the total number of packets,
they have cumulative effects when many flows are present. Thus, studying
queuing algorithms is important and can lead to significant improvements.
- Queueing algorithms control three variables: bandwidth (which
packets get transmitted), promptness (how long it takes for these packets
to be transmitted), and buffer space (which packets get discarded by the
gateway). First-Come-First-Serve (FCFS) algorithms mix the three
criteria together in assigning priority based on arrival time. A fair
algorithm proposed in this paper uses a round-robin technique to achieve
maxi-min fairness.
Critique of Contribution
This paper seems to be an early paper applying fairness techniques to
queuing. It expands beyond static round-robin, and shows good simulation
results. There is some vagueness in discussing other, non-FCFS, queueing
schemes.
Open Question
What is we use a different definition of fairness? The authors claim
that the algorithm is "fair" regardless of the definition, but does that
claim still hold if the definition of fairness is based on an external
property, such as payment?