Adaptive multitasking for dataset storage

Abstract

Data is stored in one or more data storage devices, the number of active data storage devices being adaptively adjusted to maximize parallelism while minimizing device overhead. Multiple datasets are first received, these being designated for storage in multiple data storage devices. Next, a first one of the received datasets is stored in a first one of the data storage devices. Concurrently with storage of the first dataset, the following steps are performed: (1) selecting a second received dataset from those remaining to be stored, (2) selecting an available data storage device by employing a predetermined criteria, and (3) beginning storage of the second dataset in the selected data storage device. The predetermined criteria for selecting an available data storage device may involve, for example, selecting an available device based on a 'first fit' or a 'best fit' scheme. Concurrently with storage of the second dataset, storage of a third dataset starts. Likewise, the process of storing each subsequent dataset commences while the previous dataset is being stored. This technique promotes efficiency, since later-received datasets may be stored in previously used storage devices that have already finished storing earlier datasets. Accordingly, less processing overhead is occupied in activating previously dormant data storage devices.