CRITICAL REVIEW OF EXISTING CONFIGURATION MANAGEMENT METHODS

Syntax Notation One (ASN.1) and its encoding rules are adopted by ISO and CCITT for supporting all the forms of heterogeneous interconnections (SUN has its XDR, Apollo has NCA, Xerox has XSIS, and so on). An OO model for ASN.1 has been adopted. MONDEL is a specification language having several important OO characteristics. Everything in a MONDEL specification is an object. Data structures and processes are defined as objects. A bidirectional translation MONDEL-ASN.1 has actually been implemented at the University of Montreal. VDM, Z, and B-Technologies are other description techniques. The abstarction of concurrency is not addressed in Z. The static aspects are modeled as schema, whereas the dynamic aspects are modeled as operations. At least six OO extensions of Z have been proposed since 1990 (see Appendix #5). Hardware description languages (VHDL, Lustre) cover abstract levels of hardware design. Temporal descriptions and verification of temporal properties are incorporated in these languages. Many similarities of hardware and software description languages open the way of the OO generalized approach (e.g., Estelle, VHDL, CONIC, DURRA, DARWIN, OO-TECHNIQUES). Critical Review of Existing Configuration Management Methods December 1993 P. Dini, G.v. Bochmann, A. Das, R. Dssouli 5. Automatic Reconfiguration Management Automatic reconfiguration involves three main tasks: data acquisition and data representation, data interpretation and decisions, and configuration updating. Governing all of these without human interventions constitutes automatic reconfiguration management. Data acquisition means the gathering of relevant information gather on the status of the distributed system’s components (both on topology and software configuration, and resource allocation). At the level of NE management, we have previously presented vendors’ or network managers’ solutions for local aspects. Acquisition in-the-large needs accommodation with the systems’ heterogeneity [66]. Data representation helps automatic read/write procedures to be used, as well as the standardized protocols to manage them. Data interpretation is strongly related to data representation and the method used to make decisions. Interpretation is dependent on distributed systems. Decisions are dependent both on the distributed system’s behavior and the type of problem-solving methods used. Configuration updating means making changes in the current distributed system’s status in order to improve QoS, increase fault tolerance, avoid catastrophic events, to achieve better reallocation of resources, or simply to add or remove a component (either hardware or software).