Computing Power Utilization of Distributed Systems Using Distributed Compilation: A Clustered HPC Approach
Sunil Kr Singh *
Department of CSE, Bharati Vidyapeeth’s College of Engineering New Delhi, India.
Aman Madaan
Department of CSE, Indian Institute of technology (IITB) Mumbai, India.
Ankur Aggarwal
Department of CSE, Bharati Vidyapeeth’s College of Engineering New Delhi, India.
Ankur Dewan
Department of CSE, Bharati Vidyapeeth’s College of Engineering New Delhi, India.
*Author to whom correspondence should be addressed.
Abstract
To implement a farm of computer system which compiles source programs using the computers connected in the farm. A build farm can be defined as a set of computers (distributed systems) that work together to compile software. Distributed Compiling is a special type of compiling activity in which we compile the source program using several machines with the same or cross compilers to do the compilation in parallel. In this paper, we describe a system that uses the concept for reducing build times by using free cycles of idle computer systems in distributed. The challenge of distributing compilation is tackled by a distributing compiler. We use distcc for the purpose but It would be nice if distcc could automatically detect the best distribution, but it doesn't do that yet. But in our system we have added a new php script which automatically does this work. Whenever it sees that a system is overloaded or its utilization increases above a threshold value (in our case it is 65%), it automatically removes that node. In this way we can combine the computing powers of individual machines to get a more powerful machine.
Keywords: Virtualization, cloud computing, resource monitoring, service-oriented, Infrastructure As A Service (IAAS).