Dependable Computer Systems

Dependability analysis is the recent approach to performance evaluation of contemporary systems which tries to cope with new challenges that are brought with their unprecedented complexity, size and diversity. Especially in case of computer systems and networks such evaluation must be based on multidisciplinary approach to theory, technology, and maintenance of systems which operate in real (and very often unfriendly) environments. As opposed to “classic” reliability which focuses mainly on technical aspects of system functioning, dependability studies investigate the systems as multifaceted and sophisticated amalgamations of technical, information and also human resources. This monograph presents selected new developments in such areas of dependability research as mathematical models, evaluation of software, probabilistic assessment, methodologies, tools, and technologies. Intelligent and soft computing methods help to resolve fundamental problems of dependability analysis which are caused by the fact that in contemporary computer systems it is often difficult to find a relation between system elements and system events (the relation between reasons and results) and it is even more difficult to define strict mathematical models with “analytical” relationships between such phenomena.

A team of recognized experts leads the way to dependable computing systems With computers and networks pervading every aspect of daily life, there is an ever-growing demand for dependability. In this unique resource, researchers and organizations will find the tools needed to identify and engage state-of-the-art approaches used for the specification, design, and assessment of dependable computer systems. The first part of the book addresses models and paradigms of dependable computing, and the second part deals with enabling technologies and applications. Tough issues in creating dependable computing systems are also tackled, including: * Verification techniques * Model-based evaluation * Adjudication and data fusion * Robust communications primitives * Fault tolerance * Middleware * Grid security * Dependability in IBM mainframes * Embedded software * Real-time systems Each chapter of this contributed work has been authored by a recognized expert. This is an excellent textbook for graduate and advanced undergraduate students in electrical engineering, computer engineering, and computer science, as well as a must-have reference that will help engineers, programmers, and technologists develop systems that are secure and reliable.

This book analyzes the causes of failures in computing systems, their consequences, as weIl as the existing solutions to manage them. The domain is tackled in a progressive and educational manner with two objectives: 1. The mastering of the basics of dependability domain at system level, that is to say independently ofthe technology used (hardware or software) and of the domain of application. 2. The understanding of the fundamental techniques available to prevent, to remove, to tolerate, and to forecast faults in hardware and software technologies. The first objective leads to the presentation of the general problem, the fault models and degradation mechanisms wh ich are at the origin of the failures, and finally the methods and techniques which permit the faults to be prevented, removed or tolerated. This study concerns logical systems in general, independently of the hardware and software technologies put in place. This knowledge is indispensable for two reasons: • A large part of a product' s development is independent of the technological means (expression of requirements, specification and most of the design stage). Very often, the development team does not possess this basic knowledge; hence, the dependability requirements are considered uniquely during the technological implementation. Such an approach is expensive and inefficient. Indeed, the removal of a preliminary design fault can be very difficult (if possible) if this fault is detected during the product's final testing.

Enhance your hardware/software reliability Enhancement of system reliability has been a major concern of computer users and designers ¦ and this major revision of the 1982 classic meets users' continuing need for practical information on this pressing topic. Included are case studies of reliable systems from manufacturers such as Tandem, Stratus, IBM, and Digital, as well as coverage of special systems such as the Galileo Orbiter fault protection system and AT&T telephone switching processors.

Fundamentals of Dependable Computing for Software Engineers presents the essential elements of computer system dependability. The book describes a comprehensive dependability-engineering process and explains the roles of software and software engineers in computer system dependability. Readers will learn: Why dependability matters What it means for a system to be dependable How to build a dependable software system How to assess whether a software system is adequately dependable The author focuses on the actions needed to reduce the rate of failure to an acceptable level, covering material essential for engineers developing systems with extreme consequences of failure, such as safety-critical systems, security-critical systems, and critical infrastructure systems. The text explores the systems engineering aspects of dependability and provides a framework for engineers to reason and make decisions about software and its dependability. It also offers a comprehensive approach to achieve software dependability and includes a bibliography of the most relevant literature. Emphasizing the software engineering elements of dependability, this book helps software and computer engineers in fields requiring ultra-high levels of dependability, such as avionics, medical devices, automotive electronics, weapon systems, and advanced information systems, construct software systems that are dependable and within budget and time constraints.

A comprehensive collection of benchmarks for measuring dependability in hardware-software systems As computer systems have become more complex and mission-critical, it is imperative for systems engineers and researchers to have metrics for a system's dependability, reliability, availability, and serviceability. Dependability benchmarks are useful for guiding development efforts for system providers, acquisition choices of system purchasers, and evaluations of new concepts by researchers in academia and industry. This book gathers together all dependability benchmarks developed to date by industry and academia and explains the various principles and concepts of dependability benchmarking. It collects the expert knowledge of DBench, a research project funded by the European Union, and the IFIP Special Interest Group on Dependability Benchmarking, to shed light on this important area. It also provides a large panorama of examples and recommendations for defining dependability benchmarks. Dependability Benchmarking for Computer Systems includes contributions from a credible mix of industrial and academic sources: IBM, Intel, Microsoft, Sun Microsystems, Critical Software, Carnegie Mellon University, LAAS-CNRS, Technical University of Valencia, University of Coimbra, and University of Illinois. It is an invaluable resource for engineers, researchers, system vendors, system purchasers, computer industry consultants, and system integrators.

Can we rely on computers? The individual aspects of system dependability, reliability, availability, safety, and security are the factors that determine application success. To answer this question, the text explores the integration of these dependability attributes within practical working systems. This is an important new title in the series on Dependable Computing and Fault-Tolerant Systems.

This book presents selected papers from the Fifteenth International Conference on Dependability of Computer Systems (DepCoS-RELCOMEX), which illustrate the diversity of theoretical problems in analysis of performability, reliability and security of contemporary computer systems. Covering also methodologies and practical tools involved in this field, it is a valuable reference resource for scientists, researchers, practitioners and students who are dealing with these subjects. Established in 2006, DepCoS-RELCOMEX is an annual conference series organised by Wrocław University of Science and Technology. It focuses on the dependability and performability of contemporary computer systems – topics that can provide solutions to new challenges in evaluation of their reliability and efficiency. Since they are probably the most complex technical systems ever engineered by humans, the organization of modern computer systems cannot be modelled and analysed solely as structures (however complex and distributed) built only on the basis of technical resources. Instead they should be considered as a unique blend of interacting people (their needs and behaviours), networks (together with mobile properties, iCloud organisation, Internet of Everything) and a large number of users dispersed geographically and producing an unimaginable number of applications. This new, interdisciplinary approach is developing a continually increasing range of methods which apply also the latest findings in artificial intelligence (AI) and computational intelligence (CI).