Another Fuzzy Clustering Method for Virtual Machine Resource Allocation

Cloud data centers (CDCs) can help applications entirely use limited resources to enhance computing performance by allocating virtual machines (VMs). The adaptation allocation of VM resource can promote the efficient use of CDC resources. Since VM operating data are high-dimensional and sparse from the angle of the data collecting, fuzzy c-means (FCM) clustering is an appropriate machine learning solution to effectively allocate VMs. However, FCM clustering method hardly reduces the weights of noisy features precisely to zero. Meanwhile, it is as sensitive to initialization operations as others. To tackle this issues, we suggest a parallel sparse l(1)-norm FCM clustering method, called minibatch l(1)-norm fuzzy clustering (MB-l(1)-FCM). The related scheduling algorithms are efficient for allocating resources to restrict redundant resource utilization in CDCs, thus decreasing energy consumption, maintenance and running costs. Numerical results prove that the model successfully applies to VM cluster selection.