Workload management for big data analytics
Proceedings of the 2013 ACM SIGMOD International Conference on Management of Data
| Hi-index | 0.00 |
One of the desirable properties of enterprise DBMSs is the ability to manage workloads based on their importance to business. This calls for mechanisms that allow enterprise DBMSs to be able to translate the workload importance into resource allocation plans, which means workload importance must be quantitatively evaluated in order to play its role when enterprise DBMSs allocate system resources to workloads. This paper introduces a framework for implementing the translation and a process for quantifying workload importance. Under the framework, workload importance is encapsulated into an objective function that is optimized to derive resource allocation plans. A general form of workload utility functions is developed for simplifying the job of quantifying workload importance. Experiments show their effectiveness.