A method for selecting the sizes and the ordering of the extents used to construct a file, a segment, or a virtual space of a computer system (file). The general method is defined to be any function, applied to this purpose, that, in general, attaches larger extents to the larger file addresses, and for which the selection of extent sizes is determined only by the address an extent is to reside at in the file, plus any tuning parameters. The method results in files which are mostly contiguous, and that stay mostly contiguous, irrespective of any growth or shrinkage the file may be subjected to during its lifetime. High contiguity improves performance by permitting a compact file representation, and, for disc files, improves performance by permitting larger blocks of data to be moved to or from the disc device, and by minimizing head seeks. The repeat factor, along with the doubling of extent size each time the repeat factor expires, is an important innovation that greatly reduces the internal and external fragmentation normally associated with extent based systems.
A method is provided in which a volume allocation table (VAT) is constructed in a system memory on the basis of those files stored on a plurality of disks. In constructing the VAT of the system memory, contiguous (i.e., single run) files are first allocated to a first set of locations on the disks, and then non-contiguous (i.e., multiple run) files are allocated to a second set of locations on the disks. Upon building the VAT of the system memory, copies of it are written respectively to the disks so that the contiguous and non-contiguous files can be allocated to the first and second set of locations, respectively.
A method for synchronously writing data from memory to a disk drive memory device minimizes the number of actual I/O operations by writing data to disk in large chunks. The method comprises the steps of: allocating an array for saving buffer pointers; mapping the data to the disk space; allocating a buffer for each block of disk space; copying the data into the buffer; saving a pointer to the buffer in the array and then writing physically adjacent blocks to disk in one I/O operation. The method also updates and reads address blocks in the buffer pool and writes them to disk only at the end of the write request.
A disk block controller and file system support large extended files by using special logical blocks which contain a number of sequential physical disk blocks. The disk block controller includes a file creation support unit, an extended management unit, a first and second management units, and an allocation unit. The file creation support unit separates free physical disk blocks into a region with a small number of sequential empty physical disk block spaces and a region with a large number of sequential empty physical disk block spaces, and allocates these regions to different parts of the disk. The extended management unit manages a start and sequential block numbers for sequential empty physical disk blocks having specified entries and is utilized for extended files whose logical block size is an integral multiple of a physical block. The first and second management units manages start and sequential block numbers for sequential empty physical disk block spaces that are not covered by the extended management unit. The allocation unit finds sequential empty physical disk blocks whose sequential block number is greater than the number requested according to the management data provided by the second management unit and is used for allocating as many sequential empty physical blocks as requested.
A computer system allocates locations in DASD to store a specified length of data. The allocation attempts to minimize the number of I/O operations and seek time for accessing the DASD. The system records addresses and lengths of sets of contiguous available storage locations in the DASD. Then, the system searches the record to attempt to identify a set of contiguous available storage locations having the specified length and another set of contiguous available storage locations having a length which exceeds the specified length by an amount likely to accomodate most lengths of data subsequently to be stored on the DASD. If the search is successful in identifying a set of contiguous available storage locations having the specified length, the set is allocated to store the data, and if the search is not successful in identifying a set of contiguous available storage locations having the specified length but is successful in identifying another set of contiguous available storage locations having a length which exceeds the specified length by an amount likely to accomodate most subsequent lengths of data to be stored in the DASD, then the other set is allocated. To minimize radial movement of the read/write head, only sets that are located within the same extent are recorded.
A method and apparatus are provided for adaptive localization of frequently accessed, randomly addressed data in a direct access storage device (DASD) to achieve improved system access performance. At selected sampling intervals, a DASD storage controller analyzes data access patterns based on frequency of access, identifies a remapping algorithm to remap the logical groups to physical groups, and moves the physical groups according to the identified remapping algorithm. The data reordering on the DASD provides frequently accessed data in close proximity so that seek time is minimized. The adaptive data localization method periodically performed by the storage controller is transparent to the host file system. The reordering of the data on the DASD is performed during periods of low system data transfer activity.
