a hierarchical storage unit formed of a primary storage using a high-speed, small-capacity storage apparatus and a secondary storage using a low-speed, large-capacity storage apparatus, with the primary storage being used as a cache apparatus of the secondary storage,

a block information table managing unit that generates, when a data block is written in the primary storage, a block information table on a volatile memory for management, the block information table having stored therein a time stamp, and writes block information tables on the volatile memory in the primary storage in each predetermined timing for storage;

a link managing unit that generates, when the block information table is generated, a link on the volatile memory, the link having the block information table inserted in a position so as to achieve a time-stamp order and connected by pointers, and generates a link table having stored therein the pointers for management on the volatile memory;

an input/output processing unit that performs an input/output process of the hierarchical storage unit based on a mapping check upon reception of an input/output request of a data block from a higher-level apparatus;

a restore processing unit that restores the block information tables from the primary storage to the volatile memory at power-on again;

a first sort processing unit that classifies the restored block information tables into group-specific links of a plurality of time zones and then forms a group classification link by connecting group-specific links in temporal order; and

a second sort processing unit that sequentially retrieves the block information tables of the group-specific links from an earliest time stamp side for insertion in a position so as to achieve a time-stamp ascending order to reconfigure the link, wherein

for a connection change process of inserting a process target table in an arrayed link at a position so as to achieve a time-stamp order, with an LRU side of the link being taken as a front and an MRU side of the link being taken as a back, the second sort processing unit sets in advance eight connection change process cases based on combinations of three conditions of:

a first condition indicating whether a pre-insert-position table (A) is present,

a second condition indicating whether the process target table (Y) has a post-target-position table (Z), and

a third condition indicating whether a post-insert-position table (B) and the process target table (Y) are adjacent to each other, and

every time a process target table is extracted, determines three conditions to select a relevant one of the connection change process cases and perform the connection change process.