TW266357B - Data compression system - Google Patents

Abstract

An ordered storage file preferably has a variable number of records of variable size. Each record includes a block of digital data compressed, i.e. coded, by one of a selected set of compressor algorithms. The file also contains indices for allowing individual, random retrieval of the records. The file also has a set of decompression algorithms usable for decompressing the records. The source of the compressed data of each record preferably comprises an allocation unit which is a division of the original uncompressed data. The allocation units are preferably uniform in size and each contains one or more logical sectors which preferably are equal in size to the physical sectors of the storage device on which the file resides. Each allocation unit is independently compressed and decompressed using selectable algorithms in a trade off, such as between the size of the compressed data and the speed of decompression. The storage files are created by an input controller which can be a computer run process that takes the original uncompressed data, runs the data through an operator selected set of compressors, from the results selects the best compressor, and records all the compressed allocation units plus all the decompressors needed to expand the allocation units on the file. An output controller accesses the compressed data from the file and provides it to a requester transparently. In other words, the output controller provides the data to a data requester as if it had been stored on a storage device uncompressed.