ABSTRACT The concept of a service‐level agreement (SLA), a contractual arrangement between providers and customers, is firmly established in the software engineering field. However, introducing enhanced dynamism to this service contract, in one or more directions to be subsequently specified, constitutes a challenge, an important step for the Computer Engineering community, and, in our case, an anticipated contribution to the fogSLAAntillas (fogSLA) collaborative project. Consequently, this paper seeks to improve the understanding, classification, and articulation of the terminology associated with dynamic SLAs as it has evolved in the literature. The discussion extends to methodological and practical options for integrating this concept within the fogSLA project framework. On this occasion, we share our experience with dynamic SLA in the context of scheduling pods for a Kubernetes cluster. Thus, the mentioned practical work is specifically devoted to the concept of dynamicity in SLA within the Cloud‐Fog‐Edge continuum. This survey commences with a literature review and subsequently explores potential research directions applicable to the fogSLA project and to the broader Cloud, Fog, and Edge Computing communities. Some of these paths are already well defined, but would require community endorsement to attain the status of standards in the future. Therefore, we propose novel ideas and extend beyond a mere survey of the existing literature. The main contributions of this paper are, first, to align with the goal of qualitative research by deepening the understanding of the dynamic SLA topic and the potential challenges it entails, and second, to demonstrate how we have practically tackled the problem of dynamic SLAs within a project involving industrial partners, thereby indicating a real demand for this concept.

EcoIndex has been proposed to evaluate the absolute environmental performance of a given URL using a score ranging from 0 to 100 (higher is better). In this article, we revisit the calculation method of the EcoIndex metric through low-cost Machine Learning (ML) approaches. Our research aims to extend the initial idea of analytical computation, i.e., a relation (equation) between three variables, in the direction of algorithmic Machine Learning (ML) computations, allowing to treat large numbers of data, which is not the case with the current computation. For a URL, our new calculation methods mimic the initial metric and return an environmental performance score but make fewer assumptions than the initial method. We develop several ML ways, either using learning techniques (Locality Sensitive Hashing, K Nearest Neighbor) or matrix computation constitutes the paper’s first contribution. We use standard methods to keep the solutions simple and understood by the public. The second contribution corresponds to a discussion on our implementations, available on a GitHub repository. As major findings or trends of our study, we also discuss the limits of the past and new approaches in a search for new metrics regarding the environmental performance of HTTP requests admissible by the most significant number of people. Our work refers to the uses of digital technology. Therefore, explaining the environmental footprint measures with few words seems important if we want to move towards greater digital sobriety. Otherwise, we run the risk of not being followed by civil society.

In cloud computing, elasticity is defined as "the degree to which a system is able to adapt to workload changes by provisioning and de-provisioning resources in an autonomic manner, such that at each point in time the available resources match the current demand as closely as possible". Adding elasticity to HPC (High Performance Computing) clusters management systems remains challenging even if we deploy such HPC systems in today's cloud environments. This difficulty is caused by the fact that HPC jobs scheduler needs to rely on a fixed set of resources. Every change of topology (adding or removing computing resources) leads to a global restart of the HPC jobs scheduler. This phenomenon is not a major drawback because it provides a very effective way of sharing a fixed set of resources but we think that it could be complemented by a more elastic approach. Moreover, the elasticity issue should not be reduced to the scaling of resources issues. Clouds also enable access to various technologies that enhance the services offer to users. In this paper, our approach is to use containers technology to instantiate a tailored HPC environment based on the user's reservation constraints. We claim that the introduction and use of containers in HPC job schedulers allow better management of resources, in a more economical way. From the use case of SLURM, we release a methodology for 'containerization' of HPC jobs schedulers which is pervasive i.e. spreading widely throughout any layers of job schedulers. We also provide initial experiments demonstrating that our containerized SLURM system is operational and promising.

The aim of sentiment analysis, known as opinion mining, is to discover subjective information by understanding the meaning over public opinions, standpoints and attitudes from the shared text, such as consumers feedback, which focuses on automated tools. In this paper, we introduce a novel model based on a learning approach to enhance the rate of understanding and predicting the sentiment of a text. Our proposed learning approach establishes an effective penalty mechanism to map out the links between the analyzed context in which the sentiment is similar. With a gold standard corpus we released, the results obtained are better in terms of precision, recall, and computation time cost using a multithreading model.

This paper presents a new scheduling strategy proposed to optimize the scheduling of several containers submitted online by users in a private infrastructure of nodes i.e. a cloud platform. In the literature, several scheduling frameworks and studies are proposed. The majority of these works use a scheduling strategy based on one criterion, such as Spread and Bin Packing strategies. The Spread strategy consists to select the node having the least number of executed containers to balance the containers load between all nodes of the infrastructure. The Bin Packing strategy consists to select the most compacted node in terms of resources to reduce the number of used nodes of the infrastructure. However, the submitted containers are defined according to a multi-criteria, such as the number of used CPUs and the used memory size. The state of each node is also defined according to a multi-criteria, such as the number of executed containers, the number of available CPUs and the size of available memory. The novelty of our scheduling strategy is to choose the node that executes a container by combining the Spread and the Bin Packing principles using the Technique for the Order of Prioritisation by Similarity to Ideal Solution (TOPSIS) algorithm. TOPSIS is a multi-criteria decision analysis algorithm. Our proposed strategy is implemented in Docker Swarm. Docker swarm is an important container scheduler framework developed by Docker. Experiments demonstrate the potential of our strategy under different scenarios.

Actually, recommendation systems are widely used across the internet to assist users in finding products or services that fit their individual preferences. In this paper we present a new profile recommendation approach based on Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) algorithm. TOPSIS is a multi-criteria decision analysis algorithm. Our approach can be used in the context of flatsharing between persons. The goal is to suggest to a new user profile a set of room-mates profiles that are similar to his. Our approach is proposed to improve the relation between room-mates. In our context, we suppose that we have a set of room-mates profiles saved in a database. Each profile is defined according to its weight in a quantitative multi-criteria. The principle consist to recommend profiles saved in the database according to their similarity with the user profile. In our case, the similarity between profiles is defined as a minimization and/or maximization of distance between multi-criteria. The novelty of our approach is to recommend for each new user profile a set of similar profiles according to a good compromise between the multi-criteria distance. Experiments demonstrate the potential of our approach under different scenarios.

Smart buildings present challenging opportunities and issues regarding the intelligence we push in the framework that manage the building. Among them the scheduling and allocation of tasks/jobs. Based on the expertises of Qarnot Computing and LIPN Laboratory, we address three relevant matters of concern for smart buildings. First of all, we explain the economic model that the Qarnot middleware implements for managing, in a distributed way, smart buildings. The key idea is to consider the building as a data-center, managed by a cloud middleware. Then we introduce the key properties we consider as important in the context of the construction of a middleware for buildings. At last, this paper presents an efficient scheduling and allocation system according to an innovative economic model. The idea is to presents a new container scheduling system based on SLA (Service Level Agreements) classes and which is used in smart building with cloud computing environment. The novelty of our system is based on the possibility to adapt, dynamically, the scheduling and the resources allocation of containers according to the different SLA classes and the activities peaks of the nodes in the cloud i.e. the building. Experimental results show that our system gives expected results for our scenario and provides with good performance regarding the balance between objectives.

This paper presents a hybrid partitioning method used to improve the performance of solving a Satisfiability (SAT) problem. The principle of our approach consist firstly to apply a static partitioning to decompose the search tree in finite set of disjoint sub-trees, than assign each sub-tree to one computing core. However it is not easy to choose the relevant branching variables to partition the search tree. We propose in this context to partition the search tree according to the variables that occur more frequently then others. The advantage of this method is that it gives a good disjoint sub-trees. However, the drawback is the imbalance load between all computing cores of the system. To overcome this drawback, we propose as novelty to extend the static partitioning by combining with a new dynamic partitioning that assures a good load balancing between cores. Each time a new waiting core is detected, the dynamic partitioning selects automatically using an estimation function the computing core which has the most work to do in order to partition dynamically its sub-tree in two parts. It keeps one part and gives the second part to the waiting core. Preliminary result show that a good speedup is achieved using our hybrid method.

Summary This paper describes a learning parallel constraint programming (CP) solver designed for solving CP problems with several instances on massively parallel computing platforms comprising multi‐core parallel machines or Many Integrated Cores. The CP solver proposed in this work is based on aPortfolioparallelization that employs a linear reward inaction learning algorithm in order to obtain the best possible performance for a large set of instances of the same problem. The linear reward inaction algorithm enables the prediction of the number of cores to be assigned to each search strategy based on previous experiments, reducing the computing time required to solve constraint satisfaction and optimization problems. The underlying principle of thePortfolioapproach is to runNsequential search strategies usingNcomputing cores (NToNPortfolio) where each core uses its own strategy in order to perform a search that is different from strategies used by the other cores. The first strategy that finds a solution stops all other strategies. The problem with theNToNPortfolioapproach is that the number of search strategies is very small compared with the current number of computing cores used by the parallel machines. However, using an internal parallelization for each search strategy, it is possible to runNparallel search usingPcomputing cores withP≫N(NToPPortfolio). ThisNToPPortfolioperforms suboptimally for solving different CP problems because many computing resources are wasted. To improve thisPortfoliomodel, an adaptiveNToPPortfoliowas proposed, which tries to privilege the strategy that is most likely to find a solution first in order to give it more computing cores than the other strategies. However, the main problem with the adaptivePortfoliois that it loses all the learned information at the end of each search; it is designed to solve just one CP problem. Furthermore, many computational resources are wasted by bothPortfoliosolvers, the non‐adaptive (NToNandNToP) and the adaptiveNToPPortfolio, when employed in some industrial projects, such as thePAJEROproject, which always solves different instances of the same CP problem. To minimize the amount of wasted resources and to learn the most efficient search strategies, we propose a new learningPortfoliosolver that uses a learning algorithm that configures automatically number of cores to each search. The performance obtained using the differentPortfoliosolvers is compared and illustrated by solving CP problems using as example the Google OR‐Tools solver. Copyright © 2016 John Wiley & Sons, Ltd.