Reviewer: Christoph Jechlitschek
Date: 10-6-2005
How would you rate this paper, relative to others we have read? bottom 50%
How would you rate your knowledge of the topic of this paper? novice
What problem or issue does the paper address? Why is it important?
The authors propose the use of constraint sets to evaluate the performance of different router architectures. They use an abstract model called single-buffered routers. This topic is important because routers grow bigger and bigger and memory bandwidth is one of the key restrictions.
What are the main contributions of the paper and why are they important?
The use of constraint sets to characterize the performance of routers,
and an abstract model that fits most router architectures.
How significant are these contributions relative to previous work?
I thought that it is already known, which types of routers perform best and therefore I can't judge how significant their work really is. Their main metric is memory bandwith but I'm not sure if you can reduce the performance of a router to it's memory bandwidth. I guess if we could then there would be only a small number of routers
Give detailed comments justifying your view of the paper.
This paper looks to me like a collections a different things. I couldn't form a big picture where everything fits together. The format of the paper is also strange. It ends without conclusion, and a single open problem is presented that basically says that their router would work better than other routers if it would not need global knowledge and a central scheduler. That statement alone would make me reject the paper. I guess why a PIFO DSM router has yet to be built.