In all-IP networks where all types of services are integrated using IP technology, it is highly likely that traffic congestion in one service causes congestion in other services, or that abnormal traffic in one service degrades the quality of other services. This paper proposes a flexible reallocation of both computing and network resource between services, in order to reduce the chance of congestion in an all-IP network, assuming that both computing and network resources are simultaneously allocated to each service request. This paper clarifies the method of determining whether resource reallocation is possible and, if so, the optimal amount of resources to be reallocated. It is demonstrated by simulation evaluations that the proposed reallocation scheme can avoid congestion even in cases where the number of service requests received by the network greatly exceeds the number at which congestion occurs.
展开▼
