Abstract: Nested loops are one of the most time-consuming parts and the largest sources of parallelism in many scientific applications. In this paper, we address the problem of 3-dimensional tiling and scheduling of three-level perfectly nested loops with dependencies on heterogeneous systems. To exploit the parallelism, we tile and schedule nested loops with dependencies by awareness of computational power of the processing nodes and execute them in pipeline mode. The tile size plays an important role to improve the parallel execution time of nested loops. We develop and evaluate a theoretical model to estimate the parallel execution time of tilled nested loops. Also, we propose a tiling genetic algorithm that used the proposed model to find the near-optimal tile size, minimizing the parallel execution time of dependence nested loops. We demonstrate the accuracy of theoretical model and effectiveness of the proposed tiling genetic algorithm by several experiments on heterogeneous systems. The 3D tiling reduces the parallel execution time by a factor of 1.2x to 2x over the 2D tiling, while parallelizing 3D heat equation as a benchmark.

Abstract: Virtualization represents an essential technology in Cloud computing, which allows virtual machines (VM) to be executed within their own environment on top of physical hardware. The modern methods for software delivery are utilizing the concept of Vitual Machine as a efficient tool for software packaging. Typically, VMs are created using specific templates that are stored in proprietary repositories, thus leading to provider lock-in and reduced portability in the cases of simultaneous usage of multiple federated Clouds. Unfortunately, the current state-of-the-art does not provide any efficient means for streamlined management of VM images across multiple repositories, especially within federated Cloud environments. In this paper we present a novel multi-objective middleware for distributed VMI repositories in federated Cloud environment. The middleware has been designed to provide easy to use interface capable of receiving unmodified and functionally complete VM images from its users, and transparently distribute them to a specific Cloud infrastructure in a federation with respect to their size, configuration, and geographical distribution, such that they are loaded, delivered, and executed faster and with improved QoS compared to their current behaviour.

Abstract: Latest advances in information technology and the widespread growth in different areas are producing large amounts of data. Consequently, in the past decade a large number of distributed platforms for storing and processing large datasets have been proposed. Whether in development or in production, monitoring applications running on these platforms is not an easy task and dedicated tools and platforms were proposed for this scope. In this paper we present a distributed, scalable, highly available platform able to collect, store, query and process monitoring data obtained from multiple Big Data frameworks. We present its architecture and initial results obtained.

Abstract: Smart systems in telemedicine frequently use intelligent sensor devices at large scale. Practitioners can monitor non-stop the vital parameters of hundreds of patients in real-time. The most important pillars of remote patient monitoring services are communication and data processing. Large scale data processing is done mainly using work flows. Some work flows are working in real-time, more complex ones are running for days or even for weeks on parallel and distributed infrastructures such as HPC systems and cloud. In HPC environment high number of failures can arise during health care smart systems work ow enactment, so the use of fault tolerance techniques is unavoidable. The most frequently used fault tolerance technique is checkpointing. The effectiveness of the checkpointing method depends on the checkpointing interval. In this work we give a brief overview of the different checkpointing techniques and propose two new provenance based checkpointing algorithms which uses the information stored in the work ow structure to dynamically change the frequency of checkpointing and can be efficiently used for dynamic health care smart systems.

Abstract: Over recent decades computer and software systems become more and more complex because of the applications and user requirements. The complexity makes the software systems more vulnerable to the error and bugs. Also, environmental situations affect software systems which do not react to the environmental activities. Self-healing architectures have been proposed in order to make systems defeat these problems and to make systems capable of reacting to the environmental activity. Hence, these architectures help system to become dynamic and more robust, but finding a proper architecture which can support and cover system requirements is an issue. This is particularly true in industrial environments, which consist of some known and some unknown parameters. This paper presents an architecture that can be used in some industrial environment to facilitate the process of adapting the system to unpredicted situations. This architecture has been developed over the base of RAINBOW infrastructure and it is compliant to the MAPE control loop (Autonomic Computing control loop). The paper reports also about the practical experience of implementing this architecture for a painter robot in an automotive factory, which deals with problems in painted part by itself. The proposed architecture uses rule-based reasoning and it actualizes the method of environmental modeling by using a rule-based system as the model extractor. The results of the implementation shows huge benefits in reusability and even in the quality of painting process.

Abstract: EEG processing is generally acknowledged as a computationally very intensive task. The execution of pre-processing steps, frequency domain operations and source localisation algorithms result in long execution times, which prohibit the use of high-resolution EEG brain imaging techniques outside research laboratory settings. We present a novel GPU-based streaming architecture, which has the potential to drastically reduce execution times and, at the same time, provide simultaneous 2D and 3D visualization facilities. The system uses a highly-optimised and re-configurable pipeline of CPU and GPU cores that attempts to exploit the tremendous computing power whenever possible. The system can process live data arriving from an EEG device or data stored in EEG data files. The computer drives a large display wall system consisting of four 46-inch monitors, which provides a 4K-resolution drawing surface for visualising raw EEG data, potential maps and various 3D views of the patient head. Two example brain imaging algorithms, the surface Laplacian and the spherical forward solution are used as an illustration for the effective use of the massively parallel GPU hardware in speeding up computations. The paper describes the architecture of the system, the key design decisions, and the performance optimization steps that were required to achieve sub-millisecond per-sample execution times. The control ow of the system is expressed in a very modular fashion in Java but the performance-critical algorithms are programmed in CUDA and run on the GPU. Relying on the CUDA-OpenGL interoperability bridge, the computing subsystem feeds visualisation results directly into the OpenGL pipeline, eliminating unnecessary GPU-Host data transfers. The system demonstrates that up to three orders of magnitude speedups are achievable compared to MATLAB implementations, and this processing speed can be maintained during simultaneous interactive 3D visualisation of the results.

Abstract: Ensuring the correctness of middleware that ensures interoperability of various medical devices is one of the biggest challenges in the e-health domain. Traditionally, these Device Interoperability Middleware (DIM) are analyzed using software testing. However, given the inherent incompleteness of testing and the randomness of the user behaviours, the analysis results are not guaranteed to be accurate. Some of these inaccuracies in analysis results could even put human life at risk. In order to overcome these limitations, we propose to use a probabilistic model checker PRISM for analyzing DIM. The proposed approach allows us to rigorously verify reliability properties of the given DIM and thus allows the designers to make appropriate measures to design more reliable systems. For illustration, we formally analyze a middleware that uses the HL7 FHIR and ontology-based description of the devices and a communication protocol to bridge the gap in heterogeneity for dealing with different vendors and incompatible data formats.

Abstract: Hyperspectral sensors acquire images with hundreds of spectral channels. These images have a lot of information in both spectral and spatial domain, and with this kind of information different research studies can be accomplished. In this work, we present several optimizations for hyperspectral image processing algorithms intended to detect targets in hyperspectral images. The hyperspectral image selected for our study was collected by the NASAs Airborne Visible Infra-Red Imaging Spectrometer (AVIRIS) over the World Trade Center (WTC) in New York, five days after September 11th attack. The algorithm used in our experiments is the automated target generation process (ATGP) and our optimizations comprise parallel versions of the algorithm developed using open multi-processing (OpenMP) and message passing interface (MPI). Our experiments indicate that the ATGP can be successfully implemented in parallel in multicore and cluster computing architectures, including Intel Xeon Phi.

Abstract: In a general purpose cloud system efficiencies are yet to be had from supporting diverse applications and their requirements within a storage system used for a private cloud. Supporting such diverse requirements poses a significant challenge in a storage system that supports fine grained configuration on a variety of parameters. This paper uses the Ceph distributed file system, and in particular its global parameters, to show how a single changed parameter can effect the performance for a range of access patterns when tested with an OpenStack cloud system.

Abstract: Distributed research infrastructures embrace resources, services, and users. From those, services display the highest churn rate. The reasons are twofold. First, new versions or new services emerge, gain popularity, and have to be provisioned by infrastructure operators. Secondly, the demand to make research reproducible inclines the users to share the software they used to obtain their results. Coming to grips with these high churn rates is not easy, but has the potential to speed-up scientific discovery. In this paper we will describe an extensible Software-as-a-Service (SaaS) solution conceived to facilitate seamless exchange, evolution, and deployment of software services in distributed environments. In opposite to ordinary SaaS solutions, the portfolio of the software services in our approach is extensible. We implemented means for users and developers to make their services readily available in the distributed infrastructure with minimal overhead. In this process, not only the actual software is made available but also the knowledge of how to install and configure it is conveyed, thus running instances of the new service can be provided right away. We will share the experience gained during the implementation of the extensible SaaS solution for DARIAH-DE research infrastructure. It was used to provision on-demand instances of software used in digital humanities. Subsequently, the same solution was reused in a completely different context to provide on-demand instances of UNICORE Grid middleware for training and testing purposes. The operation of our extensible SaaS, yielded additional requirements and extensions. In particular, the need for monitoring and measuring the resource usage were identified and addressed. The presented insights can help to address the problem of exchange and provision of scientific software. Other distributed infrastructures can incorporate them to improve the scalability, maintainability, user-friendliness, and sustainability of their platforms.

Abstract: Process rescheduling is a known technique to face with system heterogeneity and dynamism, being especially pertinent on Bulk Synchronous Parallel (BSP) programs. These programs are organized in a set of round-based supersteps, in which the slowest process determines the moment of synchronization. This approach motivated us to develop a first model called MigBSP, which combines computation, communication and migration costs metrics for process rescheduling decisions. MigBSP originally employed an heuristic that could select either a single or a collection of process to migrate at each load balancing invocation. The first proposal is not reactive, so you should manually setup a percentage of processes to be migrated as input parameter for the load balancing model. In this work, two novel heuristics, named MigCube and MigHull, are proposed to choose the candidate processes for migration and their destination. Both heuristics consider the use of computational geometry for plotting computation, communication and migration costs metrics in a 3D graph, so both which and where load balancing questions can be answered without any user intervention. We believe that the contribution is not only in the MigBSP landscape, but also for the BSP community, who is trying to enhance performance in round-based applications in an effortless way. In addition to the description of MigCube and MigHull, this article also presents their evaluations with performance gains of up to 42% when enabling process migration over a subset of the Grid5000 infrastructure.

Abstract: Model checking has long been used as a means of verification of formal specifications. This is a verification technique of dynamic systems that explores all possible states of the system. It determines whether the given system satisfies its specification. This technique suffers from the state explosion problem when traversing all possible states of systems. Parallel and/or distributed approaches are used to cope with the state space explosion problem. In this article, we propose a synchronized parallel algorithm of exploration based on a fixed number of threads. We present many experiments for a comparison between our parallel approach and the algorithm proposed for a parallel exploration in SPIN. We show by an experimental study that our parallel approach gives encouraging results.

Abstract: Agent-based Computing addresses the challenges in managing distributed computing systems and networks through monitoring, communication, consensus-based decision-making and coordinated actuation. As a result, intelligent agents and multi-agent systems have demonstrated the capability to use intelligence, knowledge representation and reasoning, and other social metaphors like trust, game and institution, to not only address real-world problems in a human-like way but to transcend human performance. This has had a transformative impact on many application domains, particularly e-commerce, but also on planning, logistics, manufacturing, robotics, decision support, transportation, entertainment, emergency relief and disaster management, and data mining and analytics. As one of the largest and still growing research fields of Computer Science, agent-based computing today remains a unique enabler of inter-, multi- and trans-disciplinary research. This special issue provides a selection of papers concerning the state-of-the-art research in multi-agent systems. Although the idea was initiated at 17th International Conference on Principles and Practice of Multi-Agent Systems, an open call allowed any researcher working on related topics to submit a paper for review. This special issue features four articles. The article entitled Combining PosoMAS Method Content with Scrum: Agile Software Engineering for Open Self-Organising Systems by Jan-Philipp Steghoefer, Hella Seebach, Benedikt Eberhardinger, Michael Huebschmann, Wolfgang Reif proposes PosoMAS-Scrum, an agile agent-oriented software engineering process for developing large-scale open self-organising systems. The authors also demonstrate how this method can be applied in a development effort and compare it with existing approaches. The article entitled Modelling Dynamic Normative Understanding in Agent Societies by Christopher Konstantin Frantz, Martin K. Purvis, Bastin Tony Roy Savarimuthu, Mariusz Nowostawski proposes an approach to build up normative understanding from the bottom up without using any prior knowledge. Their approach combines both existing institution representations (the structure) with a norm identification process. The article entitled A Multi-Agent Approach for Trust-based Service Discovery and Selection in Social Networks by Amine Louati, Joyce El Haddad, Suzanne Pinson proposes an approach to use multi-agent systems to model the service discovery and selection process. The authors present a trust model for social networks which supports agents in determining trustworthy service providers and allows agents to reason about trust in their distributed decision-making process. The article entitled Multi-Objective Distributed Constraint Optimization using Semi-Rings by Graham Billiau, Chee Fon Chang and Aditya Ghose proposes an extended Support Based Distributed Optimization algorithm to support Multi-Objective Distributed Constraint Optimization problem. The authors also demonstrate that by building a new DCOP definition using an idempotent semiring to measure the cost/utility of a solution, this approach is able to solve multiple objectives without a total pre-order over the set of solutions. We would like to thank the editorial board of SCPE for the chance of arranging this special issue, and all the reviewers for their hard work.

Abstract: We study the problem of enforcing information flow control (IFC) in eHealth systems. IFC mechanisms allow users to control the release and propagation of sensitive information so that confidential information is not observable to unintended principals while collaborating with other legitimate principals. We describe the methodology for modelling the information flow control requirements in a hospital domain using Pravah, a parameterised lattice-based IFC framework. The key advantage of using the parameterised security class lattice is greater precision in stating policies, enhanced usability and a reduced overhead in creating security tags. We can then use type-checking to statically verify that user programs do not violate stated security policies when accessing or manipulating data records. We discuss the main issues in designing the parameterised security class lattice.

Abstract: Correctly-rounded implementations of some elementary functions are recommended by the IEEE 754-2008 standard, which aims at ensuring portable and predictable floating-point computations. Such implementations require the solving of the Table Maker's Dilemma which implies a huge amount of computation time. These computations are embarrassingly and massively parallel, but present control flow divergence which limits performance at the SIMD parallelism level, whose share in the overall performance of current and forthcoming HPC architectures is increasing. In this paper, we show that efficiently solving the Table Maker's Dilemma on various multi-core and many-core SIMD architectures (CPUs, GPUs, Intel Xeon Phi) requires to jointly handle divergence at the algorithmic, programming and hardware levels in order to scale with the number of SIMD lanes. Depending on the architecture, the performance gains can reach 10.5x over divergent code, or be constrained by different limits that we detail.

Abstract: For many years Apache Hadoop has been used as a synonym for processing data in the MapReduce fashion. However, due to the complexity of developing MapReduce applications, adoption of this paradigm in genetics has been limited. To alleviate some of the issues, we have previously developed Cloudfl ow - a high-level pipeline framework that allows users to create sophisticated biomedical pipelines using predefined code blocks while the framework automatically translates those into the MapReduce execution model. With the introduction of the YARN resource management layer, new computational processing models such as Apache Spark are now plugable into the Hadoop ecosystem. In this paper we describe the extension of Cloudfl ow to support Apache Spark without any adaptions to already implemented pipelines. The described performance evaluation demonstrates that Spark can bring an additional boost for analysing next generation sequencing (NGS) data to the field of genetics. The Cloudflow framework is open source and freely available at https://github.com/genepi/cloud flow.

Abstract: This paper describes a comprehensive sensitivity study that was performed using high performance parallel computers to understand the importance of input parameters to a transient partially saturated finite element seepage analysis for a levee with separate soil layers of sand, silty sand, and clay. Seepage flow in this paper refers to the type of flow of water that occurred through the failed levees in New Orleans, Louisiana, USA, as a result of Hurricane Katrina. The input parameters tested were saturated hydraulic conductivity, volumetric compressibility, residual moisture content, saturated moisture content, and two van Genuchten unsaturated flow parameters. The output data compiled to show the sensitivity of the input parameters were the simulation times (days) to achieve 25%, 50%, and 75% of the steady-state values of pore pressure at the toe of the levee beneath its blanket, flow rate through the landside flux section, and the level of saturation in the levee. The use of high performance parallel computers enabled the running of thousands of scenarios using different values for the input variables. A sensitivity investigation of this magnitude has not been previously performed. The results of this investigation indicated that the more sensitive soil parameters were the saturated hydraulic conductivity and the volumetric compressibility. The unsaturated van Genuchten parameters of the landside blanket had a larger than anticipated impact on the duration of time to achieve steady state. This practical example is an excellent success story for high performance computing in that running a given simulation for a couple of hours on thousands of processors in parallel replaced over a year work using a PC.

Abstract: In this paper we discuss how to combine the method content from PosoMAS, the Process for open, self-organising Multi-Agent Systems, with the agile iterative-incremental life cycle of Scrum. The result is an agile software engineering methodology tailored to open self-organising systems. We show how the methodology has been applied in a development project and discuss the lessons learned. Finally, we compare the Scrum version of PosoMAS to other agile agent-oriented software engineering methodologies and address the selection of a suitable process.

Abstract: Mobile health has recently gained a lot of attention. Biological, environmental and behavioral data collected from mobile devices can be analyzed and transmitted directly to the person, family or health professionals for immediate and individualized care. However, due to multiplicity of mobile applications and the heterogeneity of actors involved in patient care, conflicts among the privacy policies defined by the different actors can take place. Thus, we present in this paper an approach to resolve the problem of conflicting privacy policies in e-health/m-health environments using AHP (Analytic Hierarchy Process) prioritization technique. Conflicts detection and resolution are facilitated by the adoption of S4P formal privacy policy language used as a standardized language. Finally, a case study is suggested to illustrate how our solution can be applied to resolve such conflicts.

Abstract: The foraging task is one of the canonical testbeds for cooperative robotics, in which a collection of coordinated robots have to find and transport one or more objects to one or more specific storage points. Swarm robotics has been widely considered in such situations, due to its strengths such as robustness, simplicity and scalability. Typical multi-robot foraging systems currently consider tens to hundreds of agents. This paper presents a new algorithm called Energy-aware Cooperative Switching Algorithm for Foraging (EC-SAF) that manages thousands of robots. We investigate therefore the scalability of EC-SAF algorithm and the parameters that can affect energy efficiency overtime. Results indicate that EC-SAF is scalable and effective in reducing swarm energy consumption compared to an energy-aware version of the reference well-known c-marking algorithm (Ec-marking).

Abstract: The perception of lack of control over resources deployed in the cloud may represent one of the critical factors for an organization to decide to cloudify or not its own services. The flat security features offered by commercial cloud providers to every customer, from simple practitioners to managers of huge amounts of sensitive data and services, is an additional problem. In recent years, the concept of Security Service Level Agreements (Security SLAs) is assuming a key role for the secure provisioning of cloud resources and services. This paper illustrates how to develop cloud applications that deliver services covered by Security SLAs by means of the services and tools provided by the SPECS framework, developed in the context of the SPECS (Secure Provisioning of Cloud Services based on SLA Management) European Project. The whole (SPECS) application life cycle is dealt with, in order to give a comprehensive view of the different parties involved and of the processes needed to offer security guarantees on top of cloud services. The discussed development process is exemplified by means of a real-world case study consisting in a cloud application offering a secure web container service.

Abstract: Emerging challenges for scientific communities are to efficiently process big data obtained by experimentation and computational simulations. Supercomputing architectures are available to support scalable and high performant processing environment, but many of the existing algorithm implementations are still unable to cope with its architectural complexity. One approach is to have innovative technologies that effectively use these resources and also deal with geographically dispersed large datasets. Those technologies should be accessible in a way that data scientists who are running data intensive computations do not have to deal with technical intricacies of the underling execution system. Our work primarily focuses on providing data scientists with transparent access to these resources in order to easily analyze data. Impact of our work is given by describing how we enabled access to multiple high performance computing resources through an open standards-based middleware that takes advantage of a unified data management provided by the the Global Federated File System. Our architectural design and its associated implementation is validated by a usecase that requires massivley parallel DBSCAN outlier detection on a 3D point clouds dataset.

Abstract: Checkpoint/Restart or checkpointing is a fault tolerance technique which consists on taking frequent snapshots of an application, so that, in the event of a failure, the application's state can be restored and the application's execution continued without necessarily restarting it. The advent of Cloud Computing brought new challenges with regard to this technique as Fault Tolerance needs to be supplied transparently in environments running highly heterogeneous applications. In this context, we propose two new fully transparent checkpointing approaches. Both approaches use communication-induced checkpointing and guarantee a consistent view of the applications with regard to the outside world process. The first approach is uncoordinated and creates checkpoints for applications independently. The second approach is coordinated, and applications are first grouped into clusters before the checkpointing process is started. We have compared the proposed approaches with state of the art approaches. The results show that our approaches perform better when considering the communication latencies, and the overhead on the execution of the Virtual Machines.

Abstract: Blood cell separation microdevices are designed in biomedical engineering for separation of cancer cells from blood. The movement of cancer cells particles in a continuous flow microfluidic device is a challenging problem since there are several forces incorporated. For instance, forces due to inertia, gravity, buoyancy, dielectrophoresis and virtual mass are accounted for in this system. Understanding the cell particle movement and behavior at high level of abstraction is necessary in order to avoid fundamental errors in the design of systems that can make use of this behavior. In this paper we use formal analysis in order to formalize and validate the movement of microparticles under DEP forces for blood cell separation microdevice. This is achieved by modeling the dynamic behavior that can predict the trajectory of microparticles as a transition state based system. The model is used to validate the correctness of the microdevice at early stages of the design process.

Abstract: The connectivity of devices, machines and people via Cloud infrastructure can support collaborations among doctors and specialists from different medical organisations. Such collaborations may lead to data sharing and joint tasks and activities. Hence, the collaborating organisations are responsible for managing and protecting data they share. Therefore, they should define a set of access control policies regulating the exchange of data they own. However, existing Cloud services do not offer tools to analyse these policies. In this paper, we propose a Cloud Policy Verification Service (CPVS) for the analysis and the verification of access control policies specified using XACML. The analysis process detects anomalies at two policy levels: a) intra-policy: detects discrepancies between rules within a single security policy (conflicting rules and redundancies), and b) inter-policies: detects anomalies between several security policies such as inconsistency and similarity. The verification process consists in verifying the completeness property which guarantees that each access request is either accepted or denied by the access control policy. In order to demonstrate the efficiency of our method, we also provide the time and space complexities. Finally, we present the implementation of our method and demonstrate how efficiently our approach can detect policy anomalies.

Abstract: A new multi-threaded variant of the quicksort algorithm called AQsort and its C++/OpenMP implementation are presented. AQsort operates in place and was primarily designed for high-performance computing (HPC) runtime environments. It can work with multiple arrays at once; such a functionality is frequently required in HPC and cannot be accomplished with standard C pointer-based or C++ iterator-based approach. An extensive study is provided that evaluates AQsort experimentally and compares its performance with modern multi-threaded implementations of in-place and out-of-place sorting algorithms based on OpenMP, Cilk Plus, and Intel TBB. The measurements were conducted on several leading-edge HPC architectures, namely Cray XE6 nodes with AMD Bulldozer CPUs, Cray XC40 nodes with Intel Hasswell CPUs, IBM BlueGene/Q nodes, and Intel Xeon Phi coprocessors. The obtained results show that AQsort provides good scalability and sorting performance generally comparable to its competitors. In particular cases, the performance of AQsort may be slightly lower, which is the price for its universality and ability to work with substantially larger amounts of data.

Abstract: Extreme Data is an incarnation of Big Data concept distinguished by the massive amounts of data that must be queried, communicated and analyzed in near real-time by using a very large number of memory or storage elements and exascale computing systems. Immediate examples are the scientific data produced at a rate of hundreds of gigabits-per-second that must be stored, filtered and analyzed, the millions of images per day that must be analyzed in parallel, the one billion of social data posts queried in real-time on an in-memory components database. Traditional disks or commercial storage nowadays cannot handle the extreme scale of such application data. Following the need of improvement of current concepts and technologies, we focus in this paper on the needs of data intensive applications running on systems composed of up to millions of computing elements (exascale systems). We propose in this paper a methodology to advance the state-of-the-art. The starting point is the definition of new programming paradigms, APIs, runtime tools and methodologies for expressing data-intensive tasks on exascale systems. This will pave the way for the exploitation of massive parallelism over a simplified model of the system architecture, thus promoting high performance and efficiency, offering powerful operations and mechanisms for processing extreme data sources at high speed and/or real time.

Abstract: Cloud, fog and dew computing concepts offer elastic resources that can serve scalable services. These resources can be scaled horizontally or vertically. The former is more powerful, which increases the number of same machines (scaled out) to retain the performance of the service. However, this scaling is tightly connected with the existence of a balancer in front of the scaled resources that will balance the load among the end points. In this paper, we present a successful implementation of a scalable low-level load balancer, implemented on the network layer. The scalability is tested by a series of experiments for a small scale servers providing services in the range of dew computing services. The experiments showed that it adds small latency of several milliseconds and thus it slightly reduces the performance when the distributed system is underutilized. However, the results show that the balancer achieves even a super-linear speedup (speedup greater than the number of scaled resources) for a greater load. The paper discusses also many other benefits that the balancer provides.