EN FR
EN FR


Section: New Results

Results on Heterogeneous and dynamic software architectures

Resource Monitoring and Reservation in Heterogeneous and dynamic software architectures

Software systems are more pervasive than ever nowadays. Occasionally, applications run on top of resource-constrained devices where efficient resource management is required; hence, they must be capable of coping with such limitations. However, applications require support from the runtime environment to properly deal with resource limitations. This thesis addresses the problem of supporting resource-aware programming in execution environments. In particular, it aims at offering efficient support for collecting data about the consumption of computational resources (e.g., CPU, memory), as well as efficient mechanisms to reserve resources for specific applications. In existing solutions we find two important drawbacks. First, they impose performance overhead on the execution of applications. Second, creating resource management tools for these abstractions is still a daunting task. The outcomes of this work [12] are three contributions:

  • An optimistic resource monitoring framework that reduces the cost of collecting resource consumption data.

  • A methodology to select components' bindings at deployment time in order to perform resource reservation.

  • A language to build customized memory profilers that can be used both during applications' development, and also in a production environment.

Dynamic Reasoning on Heterogeneous and dynamic software architectures

Multi-Objective Evolutionary Algorithms (MOEAs) have been successfully used to optimize various domains such as finance, science, engineering, logistics and software engineering. Nevertheless, MOEAs are still very complex to apply and require detailed knowledge about problem encoding and mutation operators to obtain an effective implementation. Software engineering paradigms such as domain-driven design aim to tackle this complexity by allowing domain experts to focus on domain logic over technical details. Similarly, in order to handle MOEA complexity, we propose an approach, using model-driven software engineering (MDE) techniques, to define fitness functions and mutation operators without MOEA encoding knowledge. Integrated into an open source modelling framework, our approach can significantly simplify development and maintenance of multi-objective optimizations. By leveraging modeling methods, our approach allows reusable optimizations and seamlessly connects MOEA and MDE paradigms. We evaluate our approach on a cloud case study and show its suitability in terms of i) complexity to implement an MOO problem, ii) complexity to adapt (maintain) this implementation caused by changes in the domain model and/or optimization goals, and iii) show that the efficiency and effectiveness of our approach [56] remains comparable to ad-hoc implementations.

A Precise Metamodel for Open Cloud Computing Interface

Open Cloud Computing Interface (OCCI) proposes one of the first widely accepted, community-based, open standards for managing any kinds of cloud resources. But as it is specified in natural language, OCCI is imprecise, ambiguous, incomplete, and needs a precise definition of its core concepts. Indeed, the OCCI Core Model has conceptual drawbacks: an imprecise semantics of its type classification system, a nonextensible data type system for OCCI attributes, a vague and limited extension concept and the absence of a configuration concept. To tackle these issues, this work proposes a precise metamodel for OCCI. This metamodel defines rigourously the static semantics of the OCCI core concepts, of a precise type classification system, of an extensible data type system, and of both extension and configuration concepts. This metamodel is based on the Eclipse Modeling Framework (EMF), its structure is encoded with Ecore and its static semantics is rigourously defined with Object Constraint Language (OCL). As a consequence, this metamodel provides a concrete language to precisely define and exchange OCCI models. The validation of our metamodel is done on the first worldwide dataset of OCCI extensions already published in the literature, and addressing inter-cloud networking, infrastructure, platform, application, service management, cloud monitoring, and autonomic computing domains, respectively. This validation highlights simplicity, consistency, correctness, completeness, and usefulness of the proposed metamodel[38] , [41] .

Using Novelty Search Approach and models@runtime for Automatic Testing Environment Setup

In search-based structural testing, metaheuristic search techniques have been frequently used to automate the test data generation. In Genetic Algorithms (GAs) for example, test data are rewarded on the basis of an objective function that represents generally the number of statements or branches covered. However, owing to the wide diversity of possible test data values, it is hard to find the set of test data that can satisfy a specific coverage criterion. In this work, we introduce the use of Novelty Search (NS) algorithm to the test data generation problem based on statement-covered criteria. We believe that such approach to test data generation is attractive because it allows the exploration of the huge space of test data within the input domain. In this approach, we seek to explore the search space without regard to any objectives. In fact, instead of having a fitness-based selection, we select test cases based on a novelty score showing how different they are compared to all other solutions evaluated so far  [47] , [48] . We also create an architecture generation framework for setup testing environment for a distributed and heterogeneous service.

Using Models@Run.time to embed an Energetic Cloud Simulator in a MAPE-K Loop

Due to high electricity consumption in the Cloud datacenters, providers aim at maximizing energy efficiency through VM consolidation, accurate resource allocation or adjusting VM usage. More generally, the provider attempts to optimize resource utilization. However, while minimizing expenses, the Cloud operator still needs to conform to SLA constraints negotiated with customers (such as latency, downtime, affinity, placement, response time or duplication). Consequently, optimizing a Cloud configuration is a multi-objective problem. As a nontrivial multi-objective optimization problem, there does not exist a single solution that simultaneously optimizes each objective. There exists a (possibly infinite) number of Pareto optimal solutions. Evolutionary algorithms are popular approaches for generating Pareto optimal solutions to a multi-objective optimization problem. Most of these solutions use a fitness function to assess the quality of the candidates. However, regarding the energy consumption estimation, the fitness function can be approximative and lead to some imprecisions compared to the real observed data. This work presents a system that uses a genetic algorithm to optimize Cloud energy consumption and machine learning techniques to improve the fitness function regarding a real distributed cluster of server. We have carried out experiments on the OpenStack platform to validate our solution. This experimentation shows that the machine learning produces an accurate energy model, predicting precise values for the simulation  [124] [40] .