Decentralized and optimal control of shared resource pools
ACM Transactions on Autonomous and Adaptive Systems (TAAS) - Special section on formal methods in pervasive computing, pervasive adaptation, and self-adaptive systems: Models and algorithms
| Hi-index | 0.00 |
Resource management is a main issue when distributed applications should meet some Quality of Service (QoS) constraints. Centralized approaches are widely used for many good reasons, but decentralized approaches are often more reliable. A local view of the system from a node is not sufficient to make sound decisions. To synchronize their actions, nodes must exchange data with each other. Exchanging data incurs a cost; hence, to reduce the amount of exchanged data is desirable. This paper presents a distributed middleware to manage resources in a decentralized manner, using an approximate scheduling and a resource-borrowing scheme to improve the total QoS of the system. On each node, managers construct a comprehensive view of resource availability in the whole system. Our borrowing scheme aims to guarantee that as many applications as possible meet their deadline, to reject as few applications as possible, while decreasing the exchange costs and increasing the use of resources. Simulations show the usefulness of the approach and give encouraging results.