CWS: a model-driven scheduling policy for correlated workloads
Proceedings of the ACM SIGMETRICS international conference on Measurement and modeling of computer systems
| Hi-index | 0.00 |
Recently, there has been an increasing interest in the researchof push-based information systems that deliver datausing broadcast in both wired and wireless environments.In push-based systems, since a server can meet all requestsfor broadcast data pending in a system with one broadcast,a larger throughput can be expected when there exists alarge number of clients. In this paper, to reduce the responsetime of data access, we propose a new schedulingstrategy of broadcast program considering an environmentwhere clients issue requests for multiple data items consecutivelywith think-time. The proposed strategy reducesthe average response time by broadcasting the optimal dataitem at present according to clients' access characteristicssuch as access frequencies, correlations among data items,and think-time.