Publisher-Centric Web Caching

EECS Joint Colloquium Distinguished Lecture Series

pic of John Chuang

Professor John Chuang
School of Information Management and Systems, U. C. Berkeley

Wednesday, November 1, 2000
Hewlett Packard Auditorium, 306 Soda Hall
4:00-5:00 p.m.


Web caches are traditionally deployed by network operators to realize bandwidth savings and latency improvements. Caches are configured and managed to maximize hit rates. From the publishers' perspective, however, cache hit rates are irrelevant. Instead, current web caches are lacking several features of value to the publishers. In the data plane, web caches cannot support dynamic generation or transformation of content. In the control plane, caches do not report usage information back to the publishers, nor allow publishers any control over the distribution and management of their objects. As a result, publishers have resorted to cache-busting and/or turned to proprietary content delivery solutions.

In this talk, I will describe Gemini, an approach to migrate from network-centric caching to publisher-centric caching over the existing, heterogeneous, global web cache infrastructure. I will discuss some of the key components of the Gemini architecture, including (i) an explicit security model to protect caches, publishers and clients from one another, (ii) a QoS framework giving publishers control over the treatment of their objects at caches, and (iii) an incremental deployment strategy to allow seamless integration with legacy caches.

This approach represents joint work with Andrew Myers and Hui Zhang of Carnegie Mellon University.


John Chuang is an Assistant Professor of Information Management and Systems at UC Berkeley. He received his Ph.D. in Engineering and Public Policy at Carnegie Mellon University, where he was also a Postdoctoral Fellow in Computer Science. His BSEE and MSEE are from USC and Stanford respectively. His research and teaching encompass the technical and economic dimensions of data networking, with particular emphasis on the infrastructural foundations that support scalable information dissemination over the Internet.