Effects of system tuning and RAM disk on the performance of a clinical laboratory information system

Improvements in the performance of a laboratory computer system do not necessarily require the replacement of major portions of the system and may not require the acquisition of any hardware at all. Major bottlenecks may exist in the ways that the operating system manages its resources and the algorithm used for timesharing decisions. Moreover, significant throughput improvements may be attainable by switching to a faster storage device if substantial disk activity is performed. In this study the fractions of time used for each of the types of tasks a laboratory computer system performs (e.g. applications programs, disk transfer, queue cycler) are defined and measured. Methods for reducing the time fractions of the various types of overhead are evaluated by doing before and after studies. The combined results of the three studies indicated that a 50% improvement could be gained through system tuning and faster storage without replacement of the computer itself.