We develop Feedback Control real-time Scheduling (FCS) as a unified framework to provide Quality of Service (QoS) guarantees in unpredictable environments (such as e-business servers on the Internet). FCS includes four major components. First, novel scheduling architectures provide performance control to a new category of QoS critical systems that cannot be addressed by traditional open loop scheduling paradigms. Second, we derive dynamic models for computing systems for the purpose of performance control. These models provide a theoretical foundation for adaptive performance control. Third, we apply established control methodology to design scheduling algorithms with proven performance guarantees, which is in contrast with existing heuristics-based solutions relying on laborious design/tuning/testing iterations. Fourth, a set of control-based performance specifications characterizes the efficiency, accuracy, and robustness of QoS guarantees. 
The generality and strength of FCS are demonstrated by its instantiations in three important applications with significantly different characteristics. First, we develop real-time CPU scheduling algorithms that guarantees low deadline miss ratios in systems where task execution times may deviate from estimations at run-time. We solve the saturation problems of real-time CPU scheduling systems with a novel integrated control structure. Second, we develop an adaptive web server architecture to provide relative and absolute delay guarantees to different service classes with unpredictable workloads. The adaptive architecture has been implemented by modifying an Apache web server. Evaluation experiments on a testbed of networked Linux PC's demonstrate that our server provides robust relative/absolute delay guarantees despite of instantaneous changes in the user population. Third, we develop a data migration executor for networked storage systems that migrate data on-line while guaranteeing specified I/O throughput of concurrent applications.

https://www.cse.wustl.edu/~lu/papers/thesis.pdf

Feedback Control Real-Time Scheduling: Framework, Modeling, and Algorithms*

A scheduling architecture for real-time control tasks is proposed. The scheduler uses feedback from execution-time measurements and feedforward from workload changes to adjust the sampling periods of the control tasks so that the combined performance of the controllers is optimized. The performance of each controller is described by a cost function. Based on the solution to the optimal resource allocation problem, explicit solutions are derived for linear and quadratic approximations of the cost functions. It is shown that a linear rescaling of the nominal sampling frequencies is optimal for both of these approximations. An extensive inverted pendulum example is presented, where the performance obtained with open-loop, feedback, combined feedback and feedforward scheduling, and earliest-deadline first scheduling are compared. The performance under earliest-deadline first scheduling is explained by studying the behavior of periodic tasks under overload conditions. It is shown that the average values of the sampling periods equal the nominal periods, rescaled by the processor utilization.

Feedback–Feedforward Scheduling of Control Tasks

Automated intrusion response is an important unsolved problem in computer security. A system called pH (for process homeostasis) is described which can successfully detect and stop intrusions before the target system is compromised. In its current form, pH monitors every executing process on a computer at the system-call level, and responds to anomalies by either delaying or aborting system calls. The paper presents the rationale for pH, its design and implementation, and a set of initial experimental results.

/pdf/automated-response-using-system-call-delays-4zq21d2q50.pdf

Automated response using system-call delays

Despite the significant body of results in real-time scheduling, many real world problems are not easily supported. While algorithms such as Earliest Deadline First, Rate Monotonic, and the Spring scheduling algorithm can support sophisticated task set characteristics (such as deadlines, precedence constraints, shared resources, jitter, etc.), they are all "open loop" scheduling algorithms. Open loop refers to the fact that once schedules are created they are not "adjusted" based on continuous feedback. While open-loop scheduling algorithms can perform well in static or dynamic systems in which the workloads can be accurately modeled, they can perform poorly in unpredictable dynamic systems. In this paper, we present a new scheduling paradigm, which we call feedback control real-time scheduling. Feedback control real-time scheduling defines error terms for schedules, monitors the error, and continuously adjusts the schedules to maintain stable performance. This paper also presents a practical feedback control real-time scheduling algorithm, FC-EDF, which is a starting point in the long-term endeavor of creating a theory and practice of feedback control scheduling.

https://www.cse.wustl.edu/~lu/papers/ecrts99.pdf

The case for feedback control real-time scheduling

Despite the significant body of results in real-time scheduling, many real world problems are not easily supported. While algorithms such as Earliest Deadline First, Rate Monotonic, and the Spring scheduling algorithm can support sophisticated task set characteristics (such as deadlines, precedence constraints, shared resources, jitter etc.), they are all "open loop" scheduling algorithms. Open loop refers to the fact that once schedules are created they are not "adjusted" based on continuous feedback. While open-loop scheduling algorithms can perform well in static or dynamic systems in which the workloads can be accurately modeled, they can perform poorly in unpredictable dynamic systems. In this paper, we present a feedback control real-time scheduling algorithm and its evaluation. Performance results demonstrate the effectiveness of the algorithm when execution times vary from the worst case and when there are major shifts of total load in the system. A key part of this feedback solution is its explicit use of deadline based metrics.

/pdf/design-and-evaluation-of-a-feedback-control-edf-scheduling-5el95nkg18.pdf

Design and evaluation of a feedback control EDF scheduling algorithm

-Real-time optimization of the overall performance of a large computer system or network such as an air-traffic control system or teleprocessing network inherently requires the introduction of adaptive control into selected control functions or sets of control functions. Such adaptive control is necessitated by the high variance of the available resource inventory and the workload resource demands within the system environment. Global management of the resulting multiloop control system becomes the responsibility of the operating system. Investigated aspects of this evolutionary extension of the operating system called the dynamically adaptive operating system (DAOS) include the general methodology and the real-time modeling and estimation of resource demands. The methodology specifies the general identification, decision, and modification processes required for an adaptive operating system. Support of such processes could be attractively provided by a peripheral or miniprocessor. The problem of estimating resource service times both with and without partial service times is explored and statistical models possessing utility for real-time modeling are developed. In particular, a dynamically partitioned second moment model (DPSMM) is described. The feasibility of such models is established with respect to a General Principle of Locality. Selected simulation results are presented to evaluate selected developments with respect to an adaptive central processing unit (CPU) scheduling discipline.

Ramamoorthy

Papers

Aspects of a Dynamically Adaptive Operating System