JMS clustering in ARJUNA message service: dependability, performance and manageability

JMS is an API specification that defines a standard way for applications to access Message-Oriented Middleware. In this short paper, we report on the ongoing work for enhancing dependability of a JMS server. We suggest that a promising approach is to use a clustered solution based on a group communication framework. The key principle that permeates the design of our clustered JMS is functional homogeneity, that is, any replica can execute part or all of any messaging task. The essential aspects we take into account when defining the clustered message server are: dependability, manageability and performance Exploiting these three aspects also leads to increased scalability and strengthens reliability by providing high levels of data redundancy. We address these requirements by exploiting three techniques, namely active replication, context awareness and the novel concept of translucent replication respectively.