Abstract

Effectiveness of message passing as the means to provide communication in distributed computing motivates the use of such strategy in highly distributed service environments such as grid and cloud computing. Grids and clouds share some important features such as multiplicity of administrative domains and complexity to create applications to be executed over these platforms. In this paper we argue that message domains can provide architectural simplicity to split distributed applications into layers. Interdependence between layers is controlled by visibility policies in each message domain. This architecture allows for the creation of infrastructures of distributed agents in which applications can be modular. I. I NTRODUCTION Highly distributed computing models such as grid and cloud computing are becoming popular as companies and academic institutions start to realize the benefits of computational resources virtualization, platform as a service, and systems as a service. But computation as a commodity comes at a price. As systems scale, so the complexity to keep them working and evolving increases. Popular systems tend to have a large number of customers, therefore, developers of those systems are expected to receive constant requests for enhancements. Although message passing is largely used to create distributed applications in a more controlled scenario, the principles of message passing can be also applied to grids and clouds. Therefore, we need to adapt system distribution to address the new questions introduced by the global scale. When computational resources are dynamically provisioned, the exchanged messages are not only about data and execution control, but may be also about resource finding, reservation, and payment. When nodes are separated by continents, multicasting must be used wisely. If public networks are utilized to transport the messages, extra care should be taken with security. If more than one organization is exchanging messages in the same grid, message exchange within nodes that belong to the same organization should be isolated from messages that provide communication between organizations. As customer expectations increase due to a high offer of remote services, so does the struggle of development teams to keep the costs of software development and software change under control. To address these problems, we advocate that global scale distributed systems should be based on reliable multicast domains. Those domains should be flexible to allow for application designers to create custom configurations that will support decisions over module partitioning, user roles, secrecy of data, and division of concerns into independent layers. The next section will present our argument for the adoption of multicast domains to address those issues. Section III presents an example of message domains. Section IV presents some details of our experience to implement the message domains using JMS. Section V presents related research. Finally, we present our conclusions in Section VI.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.