PERFORMANCE OF CLIENT-SERVER APPLICATIONS OVER CREDIT-BASED FLOW CONTROLLED ATM NETWORKS

Abstract

In credit-based flow control for ATM networks, buffer is first allocated to each VC (virtual circuit) and then credit control is applied to eachVC for avoiding possible buffer overflow. The flow control mechanism efficiently uses left over bandwidth after guaranteed traffic has been served, thus achieving excellent performance in utilization of network resources. Request/response protocol used in these applications alloww, efficient sharing of switching buffer, for the case when all the VCs sharing the same buffer are subject to the same degree of down stream congestion. The required buffer size can be as small as the minimum of bandwidth * round-trip-time (RTT) for the link and number of client's * reply size. In this dissertation, we performed a simulation to study the performance results of VC round-robin, cell FIFO, VC FIFO and VC priority scheduling policies, for the client-server application, over credit based flow controlled ATM network. We compare the performance results of VC round-robin, cell FIFO, VC FIFO scheduling policies with that of VC priority scheduling policy, for the case, when reply size varies from request to request, made by the clients. Measurements show that FIFO scheduling policy performs better than VC round-robin scheduling policy, for the case, when the reply size is same for all requests. Measurements also suggest that VC priority scheduling is more efficient then FIFO and 'VC round-robin scheduling policies, when reply size varies from request to request.