Importance

Loading Importance Indicators...

Abstract

In a one aspect, a technique for tracking the status of records in replicated databases. Sets of identifiers (e.g., record identifiers or change numbers) are stored by individual replicas, and in some cases conveyed between replicas. The sets comprise lists of intervals, with each interval being either a single number (in the case of the interval containing only one identifier) or a pair of numbers representing the low and high bounds of the interval. Stored as lists of intervals, the sets can usually be efficiently compressed for compact storage or transmission. The technique is particularly useful for tracking the status of message records in distributed messaging systems. It is also particularly useful in 'push' replication schemes as the status information represented by the sets of identifiers can be replicated simply by each replica sending its set of identifiers to other replicas. In a second aspect, a technique for handling per user read/unread data in a distributed database such as a client/server messaging system, particularly server transparent distributed databases, in which the replica server accessed by a user can change transparently to the user. Copies of the per user read/unread data are stored on a first server (e.g., the user's home server on which the user's mailbox is stored in a messaging system) and on the replica server currently accessed by the user. Changes made to the replica copy to reflect records read by the user are written back to the copy stored on the first server.

Loading the Abstract Image...

First Claim