Delta-4: A Generic Architecture for Dependable Distributed Computing

Delta-4 is a 5-nation, 13-partner project that has been investigating the achievement of dependability in open distributed systems, including real-time systems. This book describes the design and validation of the distributed fault-tolerant architecture developed within this project. The key features of the Delta-4 architecture are: (a) a distributed object-oriented application support environment; (b) built-in support for user-transparent fault­ tolerance; (c) use of multicast or group communication protocols; and (d) use of standard off­ the-shelf processors and standard local area network technology with minimum specialized hardware. The book is organized as follows: The first 3 chapters give an overview of the architecture's objectives and of the architecture itself, and compare the proposed solutions with other approaches. Chapters 4 to 12 give a more detailed insight into the Delta-4 architectural concepts. Chapters 4 and 5 are devoted to providing a firm set of general concepts and terminology regarding dependable and real-time computing. Chapter 6 is centred on fault-tolerance techniques based on distribution. The description of the architecture itself commences with a description of the Delta-4 application support environment (Deltase) in chapter 7. Two variants of the architecture - the Delta-4 Open System Architecture (OSA) and the Delta-4 Extra Performance Architecture (XPA) - are described respectively in chapters 8 and 9. Both variants of the architecture have a common underlying basis for dependable multicasting, i. e.

1. Requirements and Objectives.- 1.1. Fields of Application.- 1.2. Architectural Attributes.- 2. Overview of the Architecture.- 2.1. Dependability and Real-Time Concepts.- 2.2. Dependability and Real-Time in Delta-4.- 2.3. Architectural Sub-Systems.- 2.4. Software Environment and Management Issues.- 2.5. Validation.- 3. Comparison with other Approaches.- 3.1. Commercial Fault-Tolerance Techniques.- 3.2. Comparison of Approaches by Concept/Technique.- 3.3. Comparison of Approaches by Target System Characteristics.- 3.4. Case Studies.- 3.5. Conclusion.- 4. Dependability Concepts.- 4.1. Basic Definitions.- 4.2. On the Introduction of Dependability as a Generic Concept.- 4.3. On System Function, Behavior, Structure and Specification.- 4.4. The Impairments to Dependability.- 4.5. The Means for Dependability.- 4.6. The Attributes of Dependability.- 5. Real-Time Concepts.- 5.1. Concepts and Definitions.- 5.2. Delta-4 Approach to Real-Time.- 5.3. Real-Time Communications Protocols.- 5.4. Summary of Real-Time System Requirements.- 5.5. Conclusions.- 6. Distributed Fault-Tolerance.- 6.1. Related Work.- 6.2. Node Hardware Characteristics.- 6.3. Models of Distributed Computation.- 6.4. Replicated Software Components.- 6.5. Active Replication.- 6.6. Passive Replication.- 6.7. Semi-Active Replication.- 6.8. Group Management and Fault Treatment.- 6.9. Communications Support.- 6.10. Conclusion.- 7. Delta-4 Application Support Environment (Deltase).- 7.1. Purpose and Background.- 7.2. Computational Model.- 7.3. Engineering Model.- 7.4. Deltase Support for Distributed Fault-Tolerance.- 7.5. Engineering Support for Fault-Tolerance.- 7.6. Dependable Databases.- 8. Open System Architecture (OSA).- 8.1. Multipoint Communication System (MCS).- 8.2. System Administration.- 9. Extra Performance Architecture (XPA).- 9.1. Objectives and Definitions.- 9.2. Overview.- 9.3. Real-time and Performance.- 9.4. Dependability and Computational Models.- 9.5. Support for Distributed Real-Time Computing.- 9.6. XPA as an Integrated Machine.- 9.7. System Administration.- 10. The Atomic Multicast protocol (AMp).- 10.1. Notions about Reliable Group Communication.- 10.2. Related Work.- 10.3. System Architecture.- 10.4. Summary of xAMp Services.- 10.5. The Abstract Network.- 10.6. Two-Phase Accept Protocol.- 10.7. Performance and Real-Time.- 10.8. Implementation Issues.- 11. Fail-Silent Hardware for Distributed Systems.- 11.1. Fail-Silent Node Models.- 11.2. A Comparative Evaluation.- 11.3. Concluding Remarks.- 12. Input/Output: Interfacing the Real World.- 12.1. Inter-working with non-Delta-4 systems.- 12.2. Inter-working with the Physical Environment.- 12.3. Summary.- 13. Security.- 13.1. Principles of Intrusion-Tolerance.- 13.2. Fragmentation-Scattering Applied to File Archiving.- 13.3. Intrusion-Tolerant Security Server.- 13.4. Selection and Implementation of an Authorization Policy.- 13.5. Future Extensions.- 14. Software-Fault Tolerance.- 14.1. A Brief History — the State of the Art.- 14.2. Software-Fault Tolerance for Delta-4.- 14.3. Support Mechanisms and Features.- 14.4. Specifying Software Components for Software-fault Tolerance — Tradeoffs in an Object Environment.- 14.5. Concluding Remarks.- 15. Validation.- 15.1. Overview.- 15.2. Protocol Validation.- 15.3. Fault Injection.- 15.4. Dependability Evaluation.- 15.5. Software Reliability.- Annexes.- A. Safety-Critical and Safety-Related Systems.- B. Deterministic UNIX.- C. Statistical Effects with Schedules of many Activities.- D. Assigning Precedence Parameters.- E. Formal Failure Mode Assumptions.- F. Assumption Coverage.- G. Propagate-before-Validate Error Processing Technique.- H. Interface between XPA and OSA.- I. Timing Characteristics of Preemption Points.- J. Formal Presentation of AMp.- K. Models for the Representation of Reactive Systems.- L. Service Specifications.- M. Co-existence with non-Delta-4 Stations on the same LAN.- Glossary of Abbreviations.- References.- Author Affiliations.- The Delta-4 Consortium.