Performance issues of real time Kernels

Abstract

Real-time systems by definition are systems in which the correctness of the computed results not only depends on the results themselves, but also on the time frame in which they have been obtained. Real Time Operating Systems (RTOS) are responsible for the allocation of processors and computing resources to the collection of co-operating tasks in a way which will enable them to execute according to their timing constraints [7]. In general, real time systems are classified into hard real-time systems and soft real-time systems according to the degree of criticality in timing requirement. In hard real-time systems, time constraints as well as logical correctness must be satisfied to guarantee their correct behavior. Examples include most of the mission oriented systems for control purposes where timing errors would cause catastrophic results. On the other hand, in soft real-time systems like OLTP systems it is not always necessary to execute the transactions within time constraints. Furthermore, the hard real-time system often has distributed architecture as its inherent property. Real-time operating systems can be characterized by two requirements which are essential within its application area: timeliness and dependability [1, 2, 5].