Title:		Distribution of Stored Information in the Internet: 
======			Network Web Caching and Pushing

Speaker:	Keith W. Ross, Eurecom Institute, Sophia-Antipolis, France
========

Abstract
========

The majority of information sent through the Internet is stored information, 
including documents, images, and video. The distribution of stored 
information can be greatly facilitated by introducing into the network 
intermediate storage nodes that replicate or cache popular objects.  By 
caching popular objects close to the user, a Web cache can significantly 
reduce the number of times the same object is retransmitted over a link 
and significantly reduce the average time to retrieve an object. Further
improvements in latency performance are possible by pushing objects 
to caches and clients.

ISPs are currently adding network caches to their networks at phenomenal 
rates. Because disk capacity is growing at a rate of 60% a year, ISPs
will be able to cache terabytes of information. This points to a paradigm
shift: up to very recently the principle resource in a network was bandwidth;
now the there are two principle network resources, bandwidth and storage. 


The goal of this tutorial is twofold: First to examine in detail
caching and pushing technologies, including the challenges they will
need to meet in the upcoming years. Second, to provide an overview of 
trends and research in areas of hierarchical caching and pushing. 


Outline:
========

1. Overview of the Network Web Caching 
   - overview of caching and hierarchical caching
   - benefits of caching (including some statistics on hit rates)
   - non-transparent and transparent caching
   - problems with caching (advertisement rotation, hit reports,
cookies, etc.)
2. Case Studies
   - Caching in the backbone: NLANR caches
   - Caching at regional and institutional ISPs
   - Caching in the Europe, Asia, and South America
3. HTTP Protocol
    - HTTP/1.0
    - HTTP/1.1
    - HTTP/1.1 Cache Control
    - performance comparison of persistent and non-persistent HTTP
3. Internet Caching Protocol (ICP)
    - ICP operation
    - hierarchical caching
    - ICP message format 
    - Squid
4. Storage Management
    - replacement policies
    - garbage collection
5. Hash-Routing for Collections of Shared Web caches
    - hash routing
    - robust hash routing
    - robust hash routing for heterogeneous caches
    - performance comparison of hash routing and ICP
    - Microsoft's implementation of hash routing: CARP
6. A Survey of Caching Products
    - High-performance cache products for backbone ISPs: 
      NetCache (Network Appliance), Inktomi, Alteon, DynaCache
(InfoLibria)
    - SkyCache: Object distribution by Satellite
7. Overview of Push Technology
    - Push operation
    - Case studies: BackWeb, PointCast, Tipco, Castanet, Netscape
Netcaster,
      Microsoft Active Channels
    - Protocols: CDF and DRP
8. Overview of Some Current Research Efforts in Caching and Pushing
    - Efforts to reduce overhead traffic for cooperative caching,
including
      Cache Digests and Summary Cache
    - Efforts to combine pushing and caching, including Adaptive Caching 
      and LSAM
    - Hierarchical Caching as an Alternative to Reliable Multicast
    - Efforts to allow more objects to be cachable, including Active
Caching  
9. Overview of recent efforts to model the performance of caching
    - Zipf distribution for document popularity
    - Delay models for hierarchical caching.

Biography of Speaker
=====================

Keith  Ross is a Professor in the Multimedia Communications Department
at Institute Eurecom, in Sophia Antipolis, France.  From 1985 to December 
1997, Keith Ross was with the University of Pennsylvania as Assistant, 
Associate, and Full Professor. Keith Ross received his BS from Tufts 
University (1979) in Electrical Engineering, his MS from Columbia
University (1981) in Electrical Engineering , and his Ph.D. from the
University of Michigan (1985) in Computer, Information and Control 
Engineering. He has published over 40 papers in leading journals and has 
published a book on multiservice loss models for broadband telecommunication 
networks. He is currently writing a multimedia textbook on Internet 
protocols and data networks. He is or has been on the following editorial 
boards: Queuing Systems, Theory and Applications; Probability in the 
Engineering and Information Sciences; Operations Research; 
Telecommunications Systems; and IEEE Transactions on Automatic Control. 
He was the Program Chairman of the 1995 INFORMS Telecommunications 
Conference.

=============================== END ==================================