Adaptive model predictive control of autonomic distributed parallel computations with variable horizons and switching costs

Abstract

Autonomic computing is a paradigm for building systems capable of adapting their operation when external changes occur, such as workload variations, load surges and changes in the resource availability. The optimal configuration in terms of the number of computing resources assigned to each component must be automatically adjusted to the new environmental conditions. To accomplish the execution goals with the desired Quality of Service, decision-making strategies should be in charge of selecting the best reconfigurations by taking into account metrics like performance, efficiency (avoiding wasting resources), number and frequency of reconfigurations, and their amplitude (performing minimal modifications of the current configuration). This paper presents a decision-making strategy that merges the potential of Model Predictive Control with a cooperative optimization framework. After a description of our approach, we investigate the effect of different switching costs to model the resource allocation problem. We use a control method in which our proactive decision-making strategy (designed to use future prediction horizons) is made adaptive itself by dynamically changing the horizon length on the basis of the prediction errors. Simulations have been used to exemplify our approach and to discuss the effectiveness of the variable-horizon strategy in achieving the best trade-offs between reconfiguration metrics

A Comprehensive Survey on Parallelization and Elasticity in Stream Processing

“…e la er leads to a be er solution for the whole system than the non-cooperative approach. Additionally, Mencagli et al [101,106] examine distributed elasticity control by applying Model Predictive Control in combination with a cooperative optimization framework. In particular, the e ects of switching costs between con gurations is modeled by a mathematical function which is used by a proactive control strategy to globally optimize the elasticity decisions.…”

Stream Data Load Prediction for Resource Scaling Using Online Support Vector Regression

“…Wu et al [18] introduced Chrono Stream, which provides vertical and horizontal elasticity. Mencagli [19] proposed a multi-input, multi-output (MIMO) resource controller that automatically adapts to dynamic changes in a shared infrastructure. Such models try to estimate the complex relationship between the application performance and the resource allocation and adjust the embedded model by measuring the clients' response time.…”

Hierarchical and Frequency-Aware Model Predictive Control for Bare-Metal Cloud Applications

“…Jiao et al [18] also minimized the total cost of allocating and reconfiguring both servers and networks in two-tier data centers. De Matteis and Mencagli [19], [20] developed a strategy that controls a stream-based application in parallel distributed computing environments, and minimizes the total cost of resource allocation, reconfiguration, and penalty for service quality.…”