Skip to main content

Posts

Showing posts from June, 2012

Database Benchmarking

Benchmark actually written by a single group of programmers with essentially identical syntax for all popular RDBMS Programs and Packages. The benchmark should be run in two stages.     First Before modifications and then the benchmark can be run again after "tuning" is completed by the computer or RDBMS vendor. By comparing the both the timing i.e Before and after tuning it will be evident how much performance improvement resulted from the tuning. By comparing the original code to the modified code it will be clear how much custom programming effort is required to get the performance improvement. The benchmark should measure many different aspects of database performance for different "user levels" and database sizes. The benchmark must report the results in detail so that both the strengths and weaknesses of the DBMS are uncovered. It;s kind of SWOT analysis. The benchmark should be run in its standard form for multiple database products on a