|
Abstract : |
Computers running database management applications often manage large amounts of data. Typically, the price of the I/O sub-system is a considerable portion of the com-puting hardware. Fierce price competition demands every possible savings. Lossless data compression methods, when appropri-ately integrated with the dbms, yield sig-niflcant savings. Roughly speaking, a slight increase in cpu cycles is more than offset by savings in I/O subsystem. Various de-sign issues arise in the use of data compres-sion in the dbms- from the choice of algo-rithm, statistics collection, hardware ver-sus software based compression, location of the compression function in the overall computer system architecture, unit of com-pression, update in place, and the applica-tion of log ? to compressed data. These are methodic & y examined and trade-offs dis-cussed in the context of choices made for IBM?s DB2 dbms product. 1, |
