Turnaround Time for Messages of Differing Priorities

Queuing theory and statistical methods are used in this paper to derive formulas for determining average turnaround time in teleprocessing systems that handle messages on a priority basis. This information is needed to ensure, for example, that messages are processed within acceptable time limits and that eficient use is macle of system resources. Factors considered in arriving at the formulas described here include waiting time in message queues, message processing time, I/O waiting time, and delays for higher priority processing. Results conform closehy with those obtained from simulation studies. Average turnaround time for processing messages of different , priorities has far-reaching implications in the design of a teleprocessing system. Low average turnaround time for some types of messages may be highly desirable or even critical. For example, in an insurance application, the processing of claims, endorsements , and cancellations might each be done a t different priorities. Moreover, a claims analyst might want cIaims to be processed within five minutes, so that he could assure an insurance agent that a claimant is covered. But thought must also be given to the kind of processing required by each priority class of message. If the kind of processing is not considered, system through-put may suffer. For example, if the processing required for higher priority messages involves little I/O activity and the messages are received a t a sufficiently high rate, lower priority messages may never be processed. Thus, many of the basic decisions that must be made in designing a useful and efficient teleprocessing system require a means of determining average turnaround time for processing messages of each priority in a given installation. The purpose of this paper is to assist in determining average message turnaround time-the interval between the time that a message is received and the completion of its processing in the computer. The system assumed permits up to four separate jobs to be operating run concurrently within a single computing system having only environment