Browse Prior Art Database

Use of Intelligently Adaptive Database for Message Store in implementation of Message Products such as IBM MQSeries* Everyplace Message Queue

IP.com Disclosure Number: IPCOM000014224D
Original Publication Date: 2001-Apr-16
Included in the Prior Art Database: 2003-Jun-19
Document File: 3 page(s) / 47K

Publishing Venue

IBM

Abstract

MQSeries* Everyplace available from IBM Corporation is an MQSeries product providing messaging middleware function optimised for mobile communication.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 43% of the total text.

Page 1 of 3

  Use of Intelligently Adaptive Database for Message Store in implementation of Message Products such as IBM MQSeries* Everyplace Message Queue

MQSeries* Everyplace available from IBM Corporation is an MQSeries product providing messaging middleware function optimised for mobile communication.

     The current implementation is written in Java**, and connects laptops, servers, Personal Digital Assistants ("PDAs"), phones and unattended devices such as sensors and probes to MQSeries networks.

     In messaging products such as MQSeries Everyplace, one implementation of a MessageQueue, and derived classes uses a 3 component design as follows ...

A message acceptor. This filters incoming message according to some acceptance

criteria and forwards accepted messages to the message store. A message store. This provides storage for messages which have been accepted,

until they are either deleted from the queue or forwarded by the message dispatcher. Additionally the message store provides the ability to "browse" the set of stored messages, resulting in the selection of a (possibly empty) subset of them. Optionally the browse operation may result in the selected messages being "locked".

A message dispatcher. This forwards messages from the message store to

another message queue on another device in the network.

   The invention relates to the message store component. Current implementations of message store are as follows A memory-only implementation. This is optimised for speed, but is volatile (lost if the

device is switched off), and tends to have a small capacity, so can only store a relatively few number of messages. A disk-based file implementation. This is non-volatile and typically has a large

capacity. However this is relatively slow in operation.

     In both cases the browse function is typically implemented by a search algorithm implemented in Java. This tends to be rather slow as Java is interpreted and there may be a significant amount of file i/o overhead depending on how the indexing information for the search is stored.

     A further implementation of message store has been proposed, based on the use of a database such as IBM's DB2*. This has the advantage of having a large capacity, and highly optimised search algorithms implemented in native code rather than Java. It is hoped to delegate the search ("browse") function to the database to achieve a performance gain.

     However there are difficulties with the use of a database in the implementation of the message store. In particular it is difficult to identify the key search fields in advance, and these are required in order to organise the database effectively for efficient search.

     The invention addresses the problem of organising the database for efficient search. The invention identifies key fields dynamically in operation and makes "intelligent" decisions about performance optimising the database based on the history of previous search requests.

     In this section "message" refers only to those mes...