Control theory is emerging as a source of solutions for many problems in the computing systems domain, and in particular for resource allocation. However, the techniques proposed to date in the literature, do not successfully address the difficulty of devising allocation schemes that are general enough, provide stability guarantees, and can be parametrised and managed by system administrators, or

Motivation While the pay-as-you-go model of Infrastructure-as-a-Service (IaaS) clouds is more flexible than an in-house IT infrastructure, it still has a resource-based interface towards users, who can rent virtual computing resources over relatively long time scales. There is a fundamental mismatch between this resource-based interface and what users really care about: performance.

A number of techniques have been proposed to provide runtime performance guarantees while minimizing power consumption. One drawback of existing approaches is that they work only on a fixed set of components (or actuators) that must be specified at design time. If new components become available, these management systems must be redesigned and reimplemented. In this paper, we propose PTRADE, a nov

An important challenge in the design and implementation of self-optimizing systems is that of finding a model that maps changes in a tunable parameter (or “knob”) into an effect on the performance, power, or energy, of the overall system. This paper describes ARPE (Analyzing the Relationship between Parameters and Effectors), an open source tool to analyze the effect of parameter changes on the be

Feedback controllers are typically applied to computing systems by acting on some quantities ("tunable parameters" in the computer science lexicon, e.g., a packet drop rate) to attain some goal (e.g., a required bandwidth allocation). In other words, control loops are closed around the computing system. In the authors' opinion, this is definitely a partial use of control. Many critical parts of co

This chapter deals with Model-Based AutoTuning (MBAT). Besides describing the main flow and phases of such a tuning procedure, the main advantages of the underlying methodological approach are evidenced, namely the availability of a parametric process model to simulate the loop and assess its behaviour and characteristics such as performance and robustness, the possibility of stipulating the contr

This paper presents a library of (PID) controller models adopting the object-oriented approach, and written in Modelica. Peculiar to this work is that controllers are represented both as dynamic continuous-time and digital models, achieving consistence between the two and accounting for the functionalities of typical industrial implementations. This allows the designer to use realistic controllers

Many functions of operating systems are keen to be realised as feedback controllers. Doing so has a non negligible design impact, but also a significant payoff in terms of simplicity and generality. This paper presents a complete operating system scheduler, at present implemented in a microcontroller kernel, entirely composed of a PI-based control structure. The proposed scheduler is experimented

A recently published paper addressed the so called "PI+p" controller (i.e., a PI augmented with an additional, stable pole) and demonstrated that said controller yields interesting results particularly if tuned with the "contextual" approach, where model-based rules are used in conjunction with a relay experiment so as to provide at the same time the tuned regulator, and the tuning model. In this

In some recent papers it was shown that preemptive process schedulers in multitasking operating systems can be viewed, and above all designed, as discrete-time feedback controls with very simple (I- or PI-type) regulators, yielding significant advantages over classical scheduling policies as for time complexity and parameter interpretability. In this work, the same problem is tackled with a predic

Regulators are conceived as dynamic systems, but almost invariantly implemented as computer code. In the opinion of the authors the relationships between those two worlds are frequently overlooked in control education, although an insufficient knowledge of the involved concepts can hinder the correct operation of the realised systems. This manuscript presents a didactic activity that aims at closi

This paper investigates a novel approach to derive self-adaptive software by automatically modifying the model of the application using a control-theoretical approach. Self adaptation is achieved at the model level to assure that the model-which lives alongside the application at run-time- continues to satisfy its reliability requirements, despite changes in the environment that might lead to a vi

Self-adaptive applications are capable of changing their behavior thousands of times per second to autonomically find a way to accomplish given goals despite working in an unpredictable environment whose conditions can vary continuously. In this work we propose an evolvable system that runs self-adaptive applications on top of a heterogeneous system consisting of a general purpose processor and a

Although virtually any installed PID controller encompasses a derivative filter, and very frequently takes the standard ISA form, many tuning rules refer to the ideal controller structure, which is subsequently made realizable by introducing the so-called derivative filter a posteriori. Doing so can sometimes lead to undesired results. This paper investigates the matter and discusses two very simp

Autonomic computing systems adapt themselves thousands of times a second, to accomplish their goal despite changing environmental conditions and demands. The literature reports many decision mechanisms, but in most realizations a single one is applied. This paper compares some state-of-the-art decision making approaches, applied to a self-optimizing autonomic system that allocates resources to a s

Tuning an ideal PID and then using its parameters in a real one, can yield undesired results. Nonetheless, many rules conceived for the ideal PID are successfully used in practice, owing to their greater flexibility. A systematic method would thus be useful to extend ideal PID tuning rules to the real PID, so as to preserve said flexibility while synthesising a realistic controller. This manuscrip

This manuscript addresses the problem of process scheduling in a multitasking computing environment. The mainstream feedback-based approach to that problem preserves the existing scheduler, and adapts some of its parameters by means of convenient loops. On the contrary, in this research, the scheduler is entirely replaced by suitable control structures, synthesised and analysed in the discrete-tim

Feedback controllers are typically applied to computing systems by acting on some quantities ("tunable parameters" in the computer science lexicon, think e.g. of a packet drop rate) in order to attain some goal (e.g., a required bandwidth allocation). In other words, control loops are closed around the computing system, that exposes tunable parameters as control actuators, and allows to measure so

Many model-based tuning methods for PI(D) controllers, especially - but not exclusively - in the process control domain, employ models with delay. Doing so has advantages, basically in terms of tuning rules' simplicity, but also potential drawbacks, especially from the standpoint of performance (and sometimes robustness). This manuscript aims at initiating and fostering a critical discussion on th

A formalism was recently introduced to instrument, monitor and control computer applications based on the rate of heartbeats they emit, thereby quantitatively signaling their progress toward goals. To date, the idea was however used essentially in an heuristic manner. This work first shows that a very simple dynamic heartbeat rate model can be devised, an that said model allows to address the corr