In a distributed environment, the server responsible for the management of some resource keeps track of its state using a suitable data structure, such as a linear list. The list can be maintained either concurrently or periodically. The paper presents a fluid approximation model to analyze the performance of these two strategies. The results indicate conditions under which one strategy may be better than the other in terms of the system throughput and the variance of the service time distribution. The stability of the service system is also investigated.

The derivation of the results presents an interesting case study in probabilistic modeling. If the reader is familiar with the basics of the fluid approximation theory for queues, the mathematical treatment is easy to follow. The reader feels slightly uneasy when considering the applicability of the results, however. The authors present no validation results; not even simulation results, based on a more realistic model, are given. The reader might also want to know how this model compares with the models presented earlier. It is not self-evident that this type of analysis is feasible when the behavior of clients in a local area network is modeled.