A comparative evaluation of implementation methods for controlling the execution speed of a program by regulating I/O performance

As the performance capacity of computer hardware increases, the processing time taken by software becomes shorter and it is now possible to have programs execute complex procedures. However, processing speed is highly dependent on the actual performance of the hardware. Therefore, it is necessary to establish control mechanisms in order to execute a given set of processing instructions at a speed within the limits of the hardware capacity that matches the demands of the user as well as the nature of the service provided by the software. In addition, there are various methods for implementing such control mechanisms: in the kernel of the operating system (OS) or as a library that is external to the OS. Each method has its own advantages and disadvantages. In this paper we describe and compare different methods for implementing program execution speed regulation procedures by controlling the input-output (I/O) performance. Specifically, we describe and compare implementations of a basic method for adjusting the number of times I/O is performed and a fundamental method for adjusting the time spent on I/O. In addition, we compare an implementation of a method for regulating I/O timing implemented as a library with an implementation in the OS kernel and highlight the advantages of each ap proach.