Dynamic bandwidth provisioning for datacenter

Abstract

Today’s data centers contain thousands of computers operating a variety of service and applications. The limited network bandwidth of data centers become the performance bottleneck, or simply say that this creates a congestion due to several requests. As every customer need to access service or application at high speed. To guarantee the network performance, an efficient network bandwidth management can be leveraged, however, some research had been done so far to avoid or resolve these issues. The existing research mainly relies on a static or dynamic bandwidth allocation to virtual machines. This research paper focus on bandwidth management by allocating bandwidth to a specific service instead of allocating static or dynamic bandwidth to the entire virtual machine, where a service or application that is in the inoperative state will also be allocated bandwidth. When a client or customer is accessing his virtual machine, after authentication this request is submitted to the server that will check the details of the service or application to be provided to the client. The server will classify the service or application according to the required bandwidth, and then provides enough bandwidth to that service. As the application completes to be serviced the bandwidth will be released and be allocated to another service or application that need it to work properly. This bandwidth management method will enable the efficient sharing of bandwidth and hence improves the network performance in data centers by allocating to all the services and applications to enough bandwidth. Hence data centers will manage available limited bandwidth efficiently. KEY WORDS: Data center, Bandwidth, Clients, Users, Condor, Condor-G, service, applications, data, allocation,