Abstract:
The present day organizations and large e-business requires the computing power of servers to increase and increase. But can the service providers keep replacing with new advanced system. If they are having Real Time Cluster Management System then they can upgrade through adding-up instead of replacing the existing servers.
The Internet had in late 60's private networks, which were not interoperable. And then static web pages were published on the Internet through interoperable technologies. And then portals came and recently e-businesses. The future is going towards sharing of computing resources like SETI@home (Search for Extra-Terrestrial Intelligence) through distributing the computational power.
Suppose a Web Servers like a mail server or a search engine has the server applications run on a single computer. Now as the processing load increases there is a need to upgrade the system, thus the existing system has to be replaced, which is costly and wastage of resources. In this case you have limitation to scale the processing resource.
All the server applications are made exclusively to work on a single host processing units. Through the Cluster Management System, many instances of same code is run on different systems and thus provide scalability of processing power. This enhances the scalability and availability of mission-critical, TCP/IP-based services.
The issues needed to be taken care in this mode are very unique like, interoperability at Operating System level, Programming Language level; asynchronous messaging between hosts; abstraction layer to application and to clients; session maintenance between cluster applications and process scheduling and various other issues like availability monitoring; automatic redistribution of client traffic to the surviving hosts, remote controllability and maintaining logs of events.
This report explains one method by which Real Time Cluster Management System is achieved and how it is better with respect to other leading techniques.
V
