[Show abstract][Hide abstract] ABSTRACT: Existing off-line schedulability analysis for real-time systems can only handle periodic or sporadic tasks with known minimum inter-arrival times. Modeling sporadic tasks with fixed minimum inter-arrival times is a poor approximation for systems in which tasks arrive in bursts, but have longer intervals between the bursts. In such cases, schedulability analysis based on the existing sporadic task model is pessimistic and seriously overestimates the task's time demand. In this paper, we propose a generalized sporadic task model that characterizes arrival times more precisely than the traditional sporadic task model, and we develop a corresponding schedulability analysis that computes tighter bounds on worst-case response times. Experimental results show that when arrival time jitter increases, the new analysis more effectively guarantees schedulability of sporadic tasks.
[Show abstract][Hide abstract] ABSTRACT: Many mission-critical distributed real-time applications must handle aperiodic tasks with end-to-end deadlines. However, existing middleware (e.g., RT-CORBA) lacks schedulability analysis and run-time enforcement mecha- nisms needed to give online real-time guarantees for ape- riodic tasks. The primary contribution of this work is the design, implementation, and performance evaluation of the first realization of deferrable server and admission control mechanisms for aperiodic tasks in middleware. Empirical results on a KURT-Linux testbed demonstrate the efficiency and effectiveness of our deferrable server and admission control mechanisms in TAO's federated event service.
[Show abstract][Hide abstract] ABSTRACT: The MEAD system that we are developing employs a synergistic combination of a reactive and a proactive fault-tolerance approach in order to address unanticipated events and hazards in real-time, fault-tolerant distributed systems. The reactive fault-tolerance approach involves active monitoring of the system to adapt the provided QoS and to allocate resources based on current conditions in the system. The proactive approach involves monitoring both the distributed applications and the network to seek pre-cursors to imminent failures, and then to trigger fault-recovery mechanisms in advance of the occurrence of the failure. The underlying ideas of the MEAD system have demonstrated initial promise through our enhanced capabilities to handle failures and unanticipated events, and to reduce jitter under faulty conditions.
[Show abstract][Hide abstract] ABSTRACT: Distributed, Real-time, Embedded (DRE) systems present numerous challenges with respect to certification of their real-time behavior. Ideally, to address these we would like to build a model of our system that captures relevant information about end to end real-time requirements, resource consumptions requirements and resource availability, and subject the model to real-time scheduling analysis to predict performance. Presently, scheduling theory techniques have seen limited application in DRE systems for multiple reasons including pessimistic predictions of worst-case response times. Our study quantifies the pessimism in the predictions of worst-case response times of competing end-to-end distributed periodic tasks by comparing values observed in simulation with values computed using multiple scheduling theory techniques. Specifically we consider nongreedy synchronization protocols for tasks with a high degree of recurrence. Our results show that for an end-to-end task model nongreedy techniques, when used with proportional deadline monotonic scheduling, reduce the pessimism in worst-case response time predictions to within 5% of the actual value in over 90% of cases. These (quasi) static techniques represent a baseline against which we can evaluate emerging, control theoretic, adaptive scheduling methods.
[Show abstract][Hide abstract] ABSTRACT: Commercial off-the-shelf (COTS) middleware is now widely used to develop distributed real-time and embedded (DRE) systems. DRE systems are themselves increasingly combined to form "systems of systems" that have diverse quality of service (QoS) requirements. Earlier generations of COTS middleware, such as Object Request Brokers (ORBs) based on the CORBA 2.x standard, do not facilitate the separation of QoS policies from application functionality, which makes it hard to configure and validate complex DRE applications. The new generation of component middleware, such as the CORBA component model (CCM) based on the CORBA 3.0 standard, addresses the limitations of earlier generation middleware by establishing standards for implementing, packaging, assembling, and deploying component implementations. There has been little systematic empirical study of the performance characteristics of component middleware implementations in the context of DRE systems. This paper therefore provides three contributions to the study of CCM for DRE systems. First, we describe the challenges involved in benchmarking different CORBA component model (CCM) implementations. Second, we describe key criteria for comparing different CCM implementations using key black-box and white-box metrics. Third, we describe the design of our CCMPerf benchmarking suite to illustrate test categories that evaluate aspects of CCM implementation to determine their suitability for the DRE domain. We demonstrate CCMPerf by using it to collect metrics from a CCM implementation designed for DRE applications.
[Show abstract][Hide abstract] ABSTRACT: This paper generalizes the notion of utilization bounds for schedulability of aperiodic tasks to the case of distributed resource systems. In the basic model, aperiodically arriving tasks are processed by multiple stages of a resource pipeline within end-to-end deadlines. The authors consider a multi-dimensional space in which each dimension represents the instantaneous utilization of a single stage. A feasible region is derived in this space such that all tasks meet their deadlines as long as pipeline resource consumption remains within the feasible region. The feasible region is a multi-dimensional extension of the single-resource utilization bound giving rise to a bounding surface in the utilization space rather than a scalar bound. Extensions of the analysis are provided to non-independent tasks and arbitrary task graphs. We evaluate the performance of admission control using simulation, as well as demonstrate the applicability of these results to task schedulability analysis in the total ship computing environment envisioned by the US navy.
Preview · Article · Dec 2003 · Proceedings - International Conference on Distributed Computing Systems
[Show abstract][Hide abstract] ABSTRACT: Longevity of distributed computing middleware standards, such as CORBA, depend on their ability to support a range of applications by providing low overhead access in a uniform manner to a large variety of platforms and network capabilities. OMG's recent adaptation of Stream Control Transmission Protocol (SCTP) mapping is another instance of this trend. Applications can obtain all the benefits of this emerging protocol via a standard compliant, distributed object model. This paper reports on inte- gration of SCTP with Adaptive Communications Framework (ACE) (2) and The Ace ORB (TAO). (3) By exploiting network path multiplexing capabil- ity of SCTP we demonstrate that CORBA applications can bound the maxi- mum latencies they suffer under stressful network failures to under 50 msec.
[Show abstract][Hide abstract] ABSTRACT: Numerous technical and life cycle cost consideration trends have
led to a large increase in the use of commercial, off the shelf (COTS)
computing and networking infrastructure in military command and control
(C2) applications. Object oriented (OO) real-time middleware, such as RT
CORBA, provides the capabilities required for domain specific services
and applications. However, it is unproven whether such commercial
technologies can achieve the hard real time performance required by many
C2 system. Our study reveals that DOC middleware is suitable for some
real-time applications when appropriately coupled with operating system
and network protocol quality of service (QoS) mechanisms - areas in
which further progress is still needed