| 摘要 |
Disclosed is a disk system for controlling divided areas of a cache memory, which improves the cache hit percentage by preventing data to be accessed so frequently from being swapped out from the cache memory by data not to be accessed so frequently. Identification information is added to each I/O command issued from a CPU. The identification information denotes whether data to be accessed is user data or meta data. A disk controller, when receiving such an I/O command, selects a target virtual area from among a plurality of virtual areas set in the cache memory according to the identification information added to the received I/O command. When new data is to be stored in the cache memory upon the execution of the I/O command, the disk controller records the number of the selected virtual area in the cache memory in correspondence with the new data. Consequently, it is determined which area includes the new data stored segment. A cache data replacement is executed independently for each cache area, thereby a predetermined upper limit size of each cache memory area can be kept. |
