A Simplified Network manager for Grid and Presenting the Grid as a Computation Providing Cloud

Main Article Content

M. Sudha
M. Monica

Abstract

One of the common forms of distributed computing is grid computing. A grid uses the resources of many separate computers, loosely
connected by a network, to solve large-scale computation problems. Our approach was as follows first computationally large data is split into a
number of smaller, more manageable, working units secondly each work-unit is then sent to one member of the grid ,That member completes
processing of that work-unit in its own and sends back the result. In this architecture, there needs to be at least one host that performs the task of
assigning work-units, and then sending them, to a remote processor, as well as receive the results from remote processors. We call this unit as the
Network Manager. In addition to this assigning, sending and receiving the work-units and results, there also is the need for a host that splits tasks
into work-units and assimilates the received work units. We call this unit as the Task Broker, which we propose to design. On the server end,
there is a program for processing module, splitter and assimilator (broker). Each client has an agent running in it. Once a client is online and in
communication with server, the server sends the processing module to the client. The splitter splits data into smaller work units. Scheduler then
sends this individual work unit to client. After that, the agent in client starts up its local processing module to start processing the work unit.
Once processing of work unit is complete, it returns the result back to server. The server send received result to assimilator which combines
results from all clients for further processing.

 

Key words: Network manager, Task broker, scheduler, computational cloud

Downloads

Download data is not yet available.

Article Details

Section
Articles