Browse Prior Art Database

Multi Level Cache Consistency by Semantic Network Subscriptions

IP.com Disclosure Number: IPCOM000116926D
Original Publication Date: 1995-Nov-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 2 page(s) / 55K

Publishing Venue

IBM

Related People

Harrison, RB: AUTHOR [+2]

Abstract

In a two-level caching system, there is a cache local to each system requiring fast access to the data, and a remote cache local to the source of the data. Data distribution can occur using either a pull-model or a push-model. All the data is stored in a remote cache, and pulled to local caches who request that particular class of data.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 55% of the total text.

Multi Level Cache Consistency by Semantic Network Subscriptions

      In a two-level caching system, there is a cache local to each
system requiring fast access to the data, and a remote cache local to
the source of the data.  Data distribution can occur using either a
pull-model or a push-model.  All the data is stored in a remote
cache, and pulled to local caches who request that particular class
of data.

      A multi-level caching strategy works in a similar way, with one
or more caches in between the local and the remote caches described
above.  This is done to reduce the request load on the remote cache,
if there are many local caches requiring support.

      A problem arises in either case in managing which local caches
(or intermediate caches) have requested which classes of data, so
that when the data changes at its source, the right local caches can
be updated.

      The solution described here provides an efficient way to manage
cache consistency by implementing the cache as a semantic network.

      In a semantic network, "subjects" can be arbitrarily related by
named "relationships".  Every subject must be related by a "category"
relationship to one other subject - and these category relationships
form a tree of all the subjects in the network, headed by a root
subject called the "prime category".

      This scheme utilizes the semantic network facilities by
recording subscriptions, from local caches who have previously
requeste...