Evaluation of Congestion Control Methods for Joint Multiple Resource Allocation in Cloud Computing Environments

Abstract

As cloud computing provides not only services that have been traditionally provided on the Internet but also many other services, it has a dramatically higher risk than conventional networks that an occurrence of congestion in one service leads to congestion in other services．Unlike conventional networks, cloud computing environments should provide not only bandwidth but also processing ability simultaneously.First, this paper compares two congestion control methods (Methods A and B)in cloud computing environments, assuming that multiple types of resource are allocated simultaneously, and clarifies the effective areas of two congestion control methods with computer simulations.Method A postpones the service completion time by delaying resource allocation.Method B reduces the size of required resource and allocates to the request, extending in turn the duration of resource allocation so that the total amount of resource required by the request will be satisfied.The effective areas of two congestion control methods are clarified with computer simulations.Then, this paper compares three control methods (Methods 1, 2 and 3) tocope with the excessive generation of requests from a specific access point, which results in the degradation in service quality of requests from other access points, and clarifies the effective areas of three control methods with computer simulations.Method 1 allocates minimum resources dedicated to each access point in each center.Method 2reduces the size of required resources of requests from a specific access point, and Method 3 thins out some of requests from a specific access point.