Software Design for Resilient Computer Systems

Book Preface

Abstract This chapter introduces what design principles and application requirements we need to pursue achieving resilient functioning of computer system. Principle of simplicity, reliability, reconfigurability, scalability, and redundancy are briefly discussed. Shown that implementation of principles with architecture might include system software modification as well as redevelopment of basis hardware blocks—processing area, storage area, and interfacing area of computer system.

Nowadays, computer systems are applied in diverse and important areas such as banking, military, aviation, intensive health care, industrial control, space exploration, etc. All these areas demand highest possible reliability of functional operation, i.e., availability. Availability is defined as the readiness to provide a correct service whereas reliability is the continuity of correct service [1].

The methods to increase reliability of a system are either based on increasing the reliability of individual hardware components or the introduction of fault tolerance in the overall system design and its basic parts.

For safety critical systems, i.e., systems whose failure have disastrous consequences and possibly lead to human harm, the two main features fault tolerance (FT) and real time capability (RT) are always implicitly required and should be reflected in hardware (HW), software in general and in particular system software (SSW).

In turn, the design, development and implementation are mutually dependent processes to achieve FT and RT and should be implemented taking all possible hardware and system software resources into consideration.

The design of fault tolerant systems itself assumes that the required functionality of the applications is already known at design time, as well as potential faults and the behavior of faulty elements.

This work presents a conceptual approach to the design of an efficient on-board real time system for safety critical applications with high performance, reliability and low power consumption.

As an illustration, a design of an aircraft on-board system will be used through the whole document. This type of system depicts a typical example of a safety critical system where no faults and errors in the system can be tolerated.

We are using the standard taxonomy in dependable computing which is summarized in [1]. An error is thus a deviation from the correct system service state. The cause of such an error is called a fault and can emerge system internally or system externally.

The requirements of such an on-board system are challenging as they are almost self-contradictory: the required performance and reliability infer at first glance the need for significant power, size, weight and maintenance.

At the same time, all of them are severely limited. In terms of functionality, on-board real time systems usually implement control theory algorithms, require well-developed logic and math calculations, real-time data extraction from various external devices during operation, perform matrix calculations and do a complete prognostic of the system behavior before an event impact such as an internal flight control system error or an external plane problem becomes dangerous.