Replica determinism and flexible scheduling in hard real-time dependable systems

Abstract

AbstractÐFault-tolerant real-time systems are typically based on active replication where replicated entities are required to deliver their outputs in an identical order within a given time interval. Distributed scheduling of replicated tasks, however, violates this requirement if on-line scheduling, preemptive scheduling, or scheduling of dissimilar replicated task sets is employed. This problem of inconsistent task outputs has been solved previously by coordinating the decisions of the local schedulers such that replicated tasks are executed in an identical order. Global coordination results either in an extremely high communication effort to agree on each schedule decision or in an overly restrictive execution model where on-line scheduling, arbitrary preemptions, and nonidentically replicated task sets are not allowed. To overcome these restrictions, a new method, called timed messages, is introduced. Timed messages guarantee deterministic operation by presenting consistent message versions to the replicated tasks. This approach is based on simulated common knowledge and a sparse time base. Timed messages are very effective since they neither require communication between the local scheduler nor do they restrict usage of on-line flexible scheduling, preemptions and nonidentically replicated task sets.

Towards Real-Time Fault-Tolerant CORBA Middleware

“…In COLD PASSIVE and WARM PASSIVE systems, the recovery time needed to switch to a backup replica can be unacceptably high for DRE systems with stringent timing constraints, as shown in Section 5. Likewise, in ACTIVE replication the cost associated with providing totally ordered reliable multicast and the time needed to synchronize via proprietary group communication mechanisms can be unacceptable [35].…”

“…This experiment measures the latency experienced by the client when making invocations on an object group possessing state synchronization capabilities using the SEMI-ACTIVE replication style. We assume at this point that the ACTIVE replication without any extra protocols is not predictable, as shown in [35]. Methodology.…”

The application of statistical reliability theory in the context of Intelligent Environments: a tutorial review

“…They noted the importance of Brewer's CAP conjecture (or CAP theorem) [16,17], namely that any distributed system can at best simultaneously provide two, but not three, out of consistency (all nodes of the system can access the same data at the same time), availability (failures of individual components do not prevent the operation of remaining components) and partition (fault) tolerance (loss of an arbitrary amount of shared information does not prevent the system from operating). Poledna et al [10] investigated the predictability of the quality of service in real-time systems with replicated components and flexible scheduling schemes.…”

“…The theme of this new journal is "Reliable Intelligent Environments". However, with the exception of a few studies which have focused-in on the reliability of some individual aspects [8,10] and a number of empirical [11] or simulation [12,13] studies, very little attention appears to have been given to investigating the reliability of such IEs. Such reliability could be essential in safety-critical applications, where the well-being of people [2,3,14,15], animals and/or the natural environment [5] could be at risk in cases of the system failing.…”

Timing analysis of reliable real-time communication in CAN networks

“…In the DEAR-COTS approach, delivery delays are evaluated on a stream by stream basis, where messages are delayed accordingly to their worst-case response times, considering the case of a network disturbed by inaccessibility periods [2] [3]. It is also assumed that clocks are approximately synchronised, to guarantee the deterministic behaviour of components [9] and the correct evaluation of the delivery delays.…”