DFS-oriented persistent cache method and device across computing nodes under global aggregation namespace

Abstract

The DFS-oriented global aggregation namespace cross-computing node persistent cache method comprises the following steps: all computing nodes can access any file stored in a host computing node, a remote computing node or a data storage server on demand; the computing nodes copy or migrate relevant files from the data storage server to the corresponding local persistent cache devices of the multiple computing nodes according to the cache type requirements by using the striping technology in reverse according to the file layout information given by the metadata server and the cache pre-fetching strategy preset by the strategy engine according to the task requirements; when the data cached in the multiple computing nodes meets the preset condition, the relevant computing node end data is replaced or expelled to the data storage server end by using the cache replacement or eviction strategy preset by the strategy engine; and the relevant operation is completed in cooperation according to whether the relevant files to be operated have been cached in the local persistent storage of the corresponding computing nodes.

Images