Chapter

Pervasive Computing

Volume 2414 of the series Lecture Notes in Computer Science pp 282-296

Date:

SCAN: A Dynamic, Scalable, and Efficient Content Distribution Network

  • Yan ChenAffiliated withComputer Science Division, University of California at Berkeley
  • , Randy H. KatzAffiliated withComputer Science Division, University of California at Berkeley
  • , John D. KubiatowiczAffiliated withComputer Science Division, University of California at Berkeley

* Final gross prices may vary according to local VAT.

Get Access

Abstract

We present SCAN, the Scalable Content Access Network. SCAN combines dynamic replica placement with a self-organizing application-level multicast tree to meet client QoS and server resource constraints. It utilizes an underlying distributed object routing and location system (DOLR) as an essential component. Simulation results on both flash-crowd-like synthetic workloads and real Web server traces show that SCAN deploys close to an optimal number of replicas, achieves good load balance, and incurs a small delay and bandwidth penalty for update multicast relative to static replica placement on IP multicast. We envision that SCAN could enhance a number of different applications, such as content distribution and peer-to-peer file sharing.