The CRISP Web cache

The CRISP Web Cache is a scalable Internet proxy cache developed in a collaboration with Misha Rabinovich and others at AT&T Research. The CRISP cache is one aspect of a larger project studying Caching and Replication for Internet Service Performance at AT&T.

CRISP caches are structured as a collection of autonomous proxy servers that share their cache contents through a mapping service. The first paper (HOTOS-VI) below describes this in high level and describes our first implementation. The second paper explores several possible implementations of a mapping service, and describes Crispy Squid. The third paper (WISP98) describes Crispy Squid, our most current implementation of the CRISP Web cache architecture.

The initial prototype CRISP cache was built by Syam Gadde. AT&T is currently providing a limited release of the source code; for more information, contact Misha Rabinovich.


We are releasing Crispy Squid v1.1, our latest implementation of the CRISP web cache. This is implemented as an extension to version 2.2.STABLE5 of the Squid Internet Object Cache. This is released as a patch to Squid 2.2.STABLE5:

To apply the diffs, you must have already downloaded Squid v2.2.STABLE5 from the Squid home page. Instructions are included on how to make your Squid source tree "crispy".

This latest version includes support for Vicinity Cache.

CRISP Papers and related publications


Proxycizer is a suite of simulation tools and HTTP proxy drivers used in generating data for the papers above. Please see the Proxycizer page for more details.

Syam Gadde
Last modified: Mon Nov 15 10:45:58 EST 1999