Regional division and space distribution method in cross-domain virtual data space

Abstract

The invention provides a regional division and space distribution method in cross-domain virtual data space, capable of realizing unified management and access of global data through a cross-domain unified region division method based on a global name space formed by aggregation of decentralized autonomous storage resources, and realizing space distribution for optimizing global data placement based on a virtual view. The regional division and space distribution method comprises the following steps: firstly, formulating modeling expressions of a virtual data region and a virtual storage space,and providing a theoretical basis for regional division and space distribution; then, studying a data region divisional mechanism, realizing and recording customized data regional division in globalname space nodes to support unified management and access of global data; and finally, studying a virtual storage space distribution method, achieving distribution of virtual storage space mapped by the data region, and mapping the customized region into actual storage resources of the bottom layer. Therefore, the mapping relation between the logic data region and the actual physical resources isestablished to support the global unified scheduling of the large-scale computing application.

Description

[0001] Technical field: [0002] The invention discloses a region division and space allocation method in a cross-domain virtual data space, relates to challenges faced by wide-area high-performance computing, and belongs to the field of computer technology. [0003] Background technique: [0004] At present, storage resources in the national high-performance computing environment are widely dispersed and isolated and autonomous. Large-scale computing applications urgently need a global data space that can support cross-domain unified access. On this basis, implement a cross-domain unified area division method and a virtual view-based space allocation method, allocate and layout storage resources on demand, and realize the mapping conversion between virtual areas and physical resources, so as to shield the heterogeneity and cross-border resources of underlying resources. Domain dispersion issues, thus effectively supporting the global data management and access of storage resou...

AI Technical Summary

Problems solved by technology

[0015] Cloud computing-oriented regional division and space allocation methods loosely manage storage resources, mainly provide the use of underlying resources in the form of data centers, run decentralized applications with relatively concentrated resources, recognize the heterogeneity of resources and do not emphasize the virtualization of the upper layer. organization, there are certain limitations on global access and unified management of data, and it is not suitable for the current national high-performance computing environment

The regional division and space allocation methods for grid computing can effectively use aggregated global resources, but they need dedicated network support to improve performance, and the management of data is still loose, which is not conducive to the global unified scheduling of large-scale applications. Most of the data placement strategies are based on The data center is optimized as a unit, lacking a global unified placement strategy based on virtual views

Images