| 摘要 |
In a distributed storage system having a plurality of nodes according to the present invention, a first node of the plurality of nodes determines whether first data, which is written in a first virtual partial region managed by the first node, from among a plurality of virtual partial regions, is identical to data in virtual partial regions managed by other nodes of the plurality of nodes, and if a virtual partial region managed by any other node stores written data identical to the first data, then the first node performs inter-node deduplication whereby either the first virtual partial region is reallocated to the logical partial region to which the virtual partial region that stores the identical data is currently allocated, or the virtual partial region that stores the identical data is reallocated to the logical partial region to which the first virtual partial region is currently allocated. If the I/O load on the first node has fallen below a first threshold value after the completion of the inter-node deduplication in the first virtual partial region, then the first node performs inter-node deduplication in a second virtual partial region managed by the first node, from among the plurality of virtual partial regions, which are obtained by dividing a virtual storage region. |
