Hit Rate vs. Hit Probability Based Cache Utility Maximization

Abstract Caches play a prominent role in networks and distributed systems, and their importance is reflected in much recent work on performance analysis of caching algorithms. A plethora of research has been done on the analysis of caching algorithms using the metric of hit probability under the Independent Reference Model (IRM). However, there has been a tremendous increase in the demand of different types of content with different quality of service requirements; consequently, the user needs in the networks become more heterogeneous. In order to meet such challenges, the design and analysis of content delivery networks need to incorporate service differentiation among different classes of contents and applications. Though considerable literature has focused on the design of fair and efficient caching algorithms for content distribution, little work has focused on the provision of multiple levels of service in network and web caches.
Authors
  • Nitish Panigrahy (UMass)
  • Jian Li (UMass)
  • Don Towsley (UMass)
Date Jun-2017
Venue 2017 Workshop on MAthematical performance Modeling and Analysis