Network Interface Cards (NICs) greatly evolved from simple basic devices moving traffic in and out of the network to complex heterogeneous systems offloading host CPUs from performing complex tasks on in-transit packets. These latter comprise different types of devices, ranging from NICs accelerating fixed specific functions (e.g., on-the-fly data compression/decompression, checksum computation, data encryption, etc.) to complex Systems-on-Chip (SoC) equipped with both general purpose processors and specialized engines (Smart-NICs). Similarly, Field Programmable Gate Arrays (FPGAs) moved from pure reprogrammable devices to modern heterogeneous systems comprising general-purpose processors, real-time cores and even AI-oriented engines. Furthermore, the availability of high-speed network interfaces (e.g., SFPs) makes modern FPGAs a good choice for implementing Smart-NICs. In this work, we extended the functionalities offered by an open-source NIC implementation (Corundum) by enabling time-aware traffic management in hardware, and using this feature to control the bandwidth associated with different traffic classes. By exposing dedicated control registers on the AXI bus, the driver of the NIC can easily configure the transmission bandwidth of different prioritized queues. Basically, each control register is associated with a specific transmission queue (Corundum can expose up to thousands of transmission and receiving queues), and sets up the fraction of time in a transmission window which the queue is supposed to get access the output port and transmit the packets. Queues are then prioritized and associated to different traffic classes through the Linux QDISC mechanism. Experimental evaluation demonstrates that the approach allows to properly manage the bandwidth reserved to the different transmission flows.

Quantum machines are among the most promising technologies expected to provide significant improvements in the following years. However, bridging the gap between real-world applications and their implementation on quantum hardware is still a complicated task. One of the main challenges is to represent through qubits (i.e., the basic units of quantum information) the problems of interest. According to the specific technology under-lying the quantum machine, it is necessary to implement a proper representation strategy, generally referred to as embedding. This paper introduces a neural-enhanced optimization framework to solve the constrained unit disk problem, which arises in the context of qubits positioning for neutral atoms-based quantum hardware. The proposed approach involves a modified autoencoder model, i.e., the Distances Encoder Network, and a custom loss, i.e., the Embedding Loss Function, respectively, to compute Euclidean distances and model the optimization constraints. The core idea behind this design relies on the capability of neural networks to approximate non-linear transformations to make the Distances Encoder Network learn the spatial transformation that maps initial non-feasible solutions of the constrained unit disk problem into feasible ones. The proposed approach outperforms classical solvers, given fixed comparable computation times, and paves the way to address other optimization problems through a similar strategy.

High-Performance Computing (HPC) domain provided the necessary tools to support the scientific and industrial advancements we all have seen during the last decades. HPC is a broad domain targeting to provide both software and hardware solutions as well as envisioning methodologies that allow achieving goals of interest, such as system performance and energy efficiency. In this context, supercomputers have been the vehicle for developing and testing the most advanced technologies since their first appearance. Unlike cloud computing resources that are provided to the end-users in an on-demand fashion in the form of virtualized resources (i.e., virtual machines and containers), supercomputers’ resources are generally served through State-of-the-Art batch schedulers (e.g., SLURM, PBS, LSF, HTCondor). As such, the users submit their computational jobs to the system, which manages their execution with the support of queues. In this regard, predicting the behaviour of the jobs in the batch scheduler queues becomes worth it. Indeed, there are many cases where a deeper knowledge of the time experienced by a job in a queue (e.g., the submission of check-pointed jobs or the submission of jobs with execution dependencies) allows exploring more effective workflow orchestration policies. In this work, we focused on applying machine learning (ML) techniques to learn from the historical data collected from the queuing system of real supercomputers, aiming at predicting the time spent on a queue by a given job. Specifically, we applied both unsupervised learning (UL) and supervised learning (SL) techniques to define the most effective features for the prediction task and the actual prediction of the queue waiting time. For this purpose, two approaches have been explored: on one side, the prediction of ranges on jobs’ queuing times (classification approach) and, on the other side, the prediction of the waiting time at the minutes level (regression approach). Experimental results highlight the strong relationship between the SL models’ performances and the way the dataset is split. At the end of the prediction step, we present the uncertainty quantification approach, i.e., a tool to associate the predictions with reliability metrics, based on variance estimation.

Abstract Heterogeneous computing is the major driving factor in designing new energy-efficient high-performance computing systems. Despite the broad adoption of GPUs and other specialized architectures, the interest in spatial architectures like field-programmable gate arrays (FPGAs) has grown. While combining high performance, low power consumption and high adaptability constitute an advantage, these devices still suffer from a weak software ecosystem, which forces application developers to use tools requiring deep knowledge of the underlying system, often leaving legacy code (e.g., Fortran applications) unsupported. By realizing this, we describe a methodology for porting Fortran (legacy) code on modern FPGA architectures, with the target of preserving performance/power ratios. Aimed as an experience report, we considered an industrial computational fluid dynamics application to demonstrate that our methodology produces synthesizable OpenCL codes targeting Intel Arria10 and Stratix10 devices. Although performance gain is not far beyond that of the original CPU code (we obtained a relative speedup of $$\times$$ × 0.59 and $$\times$$ × 0.63, respectively, for a single optimized main kernel, while only on the Stratix10 we achieved $$\times$$ × 2.56 by replicating the main optimized kernel 4 times), our results are quite encouraging to drawn the path for further investigations. This paper also reports some major criticalities in porting Fortran code on FPGA architectures.

The proliferation of smart connected devices using digital assistants activated by voice commands (e.g., Apple Siri, Google Assistant, Amazon Alexa, etc.) is raising the interest in algorithms to localize and recognize audio sources. Among the others, deep neural networks (DNNs) are seen as a promising approach to accomplish such task. Unlike other approaches, DNNs can categorize received events, thus discriminating between events of interests and not even in presence of noise. Despite their advantages, DNNs require large datasets to be trained. Thus, tools for generating datasets are of great value, being able to accelerate the development of advanced learning models.

Online scheduling has been an attractive field of research for over three decades.Some recent developments suggest that Reinforcement Learning (RL) techniques have the potential to deal with online scheduling issues effectively.Driven by an industrial application, in this paper we apply four of the most important RL techniques, namely Q-learning, Sarsa, Watkins's Q(λ), and Sarsa(λ), to the online single-machine scheduling problem.Our main goal is to provide insights on how such techniques perform.The numerical results show that Watkins's Q(λ) performs best in minimizing the total tardiness of the scheduling process.

Modern Cloud architectures encompass computing and communication elements that span from traditional data center computing nodes (offering almost infinite resources to satisfy any application demands) to edge-computing and IoT devices (to sense and act on the real world). This paper presents the Cloud architecture devised within the OPERA project [1], which provides new levels of energy efficiency as a full chip-to-Cloud solution. Focusing on a smart vision application (i.e., road traffic monitoring), the paper presents novel architectural solutions optimised to achieve high energy efficiency at any level: i) the computing elements supporting the acceleration of State-of-the-Art CNNs; and ii) an innovative wireless communication subsystem. Unlike conventional designs, our wireless communication subsystem exploits the advantages of software defined radio (SDN) firmware to control a reconfigurable antenna. To further extend the application range, an energy harvesting module is used to supply power. Besides the edge-IoT, high-density accelerated servers offer capabilities of running complex algorithms within a small power envelop. The effectiveness of the whole architecture has been tested in a real context (i.e., 2 installation sites). In-field measurements demonstrate our claim: high-performance coupled with high energy efficiency over the whole system.

Biological analysis applications are usually high demanding in terms of computational power required. Cloud Computing infrastructures can be of great value supporting those type of applications, thanks to the high flexibility and performing hardware provided. The merging of solutions based on MapReduce and distributed file systems services, allows the creation of scalable infrastructures and data management services adaptable to various use cases and requirements. However, the implementation and deployment of cloud based services remains a complex task for biological researchers without specific skills. In this paper authors describe a service that allows the execution of a biological networks simulation tool in cloud, implemented with the aid of MapReduce algorithms called MaReX. The paper shows in detail the interactions occurring between the various components of the platform such as: database, process manager agent and graphical user interface.

There is a strong relationship between scientific research and technology advancement. The former generally focuses on studying phenomena happening in the real world, the latter improves tools that are at the basis of this research. From this perspective, information and communication technologies allow the implementation of ever faster tools for analyzing data generated by experiments. The aim of DemoGRAPE project is to study the interaction of the upper earth atmosphere and the GNSS (Global Navigation Satellite Systems) signals received at ground, in critical environments such as the polar regions. This paper describes the ICT infrastructure used to manage the software applications used to analyzed data collected during project experimental campaigns, taking into account the following constraints: (i) the intellectual property of the applications must be protected, (ii) the underlying infrastructure resembles a cloud-federation, (iii) data are over-sized, (iv) provide a unified vision of the available resources to the user (i.e., what are the available applications, and where experimental data reside). Leveraging on a lightweight virtualization system, we proposed a management system that copes with all these four constraints. A case study is used to show the process of deploying an application through the proposed system on a specific node where data of interest reside.

The continuous evolution of information and communication technology has led to a change in the adopted computing paradigms over time. Cloud computing is an emerging paradigm in which users, depending on their specific requirements, access to a shared pool of computing resources dynamically allocated. Cloud computing represents, with respect to Grid computing, the evolutionary step towards the implementation of a ubiquitous computing service. Such paradigm leverages on the infrastructural capabilities (compute, storage, and network) of modern data centers to provide an adequate level of computational power able to satisfy users' requests. However, trying to continuously increase such capabilities comes at the cost of an increased energy consumption. Energy efficiency is, therefore, one of the major challenges that cloud providers must address. The OPERA project aims at bringing innovative solutions to increase the energy efficiency of cloud infrastructures, by leveraging on modular, high-density, heterogeneous and low power computing systems, which are able to cover the whole computing continuum. To this end, the project will design a high-density server solution in which low power processors and FPGA devices will be used to accelerate cloud workloads. High-speed optical interconnections will be used to connect the proposed server with high-performance nodes, such as OpenPOWER-based machines. Cyber-Physical Systems (CPS) represents a natural extension of cloud infrastructures since they can collect and process data locally, more specifically where they were generated. OPERA aims at researching energy efficiency of such cloud end-nodes by designing an ultra-low power computing system with reconfigurable radio frequency capabilities. The effectiveness of the whole platform will be demonstrated with key scenarios, specifically a road traffic monitoring application, the deployment of a virtual desktop infrastructure, and the deployment of a small data center on a truck.

We present a case study on the migration to a Cloud Computing environment of the advanced differential synthetic aperture radar interferometry (DInSAR) technique, referred to as Small BAseline Subset (SBAS), which is widely used for the investigation of Earth surface deformation phenomena. In particular, we focus on the SBAS parallel algorithmic solution, namely P-SBAS, that allows the production of mean deformation velocity maps and the corresponding displacement time-series from a temporal sequence of radar images by exploiting distributed computing architectures. The Cloud migration is carried out by encapsulating the overall P-SBAS application in virtual machines running on the Cloud; moreover, the Cloud resources provisioning and configuration phases are implemented in an automatic way. Such an approach allows us to preserve the P-SBAS parallelization strategy and to straightforwardly evaluate its performance within a Cloud environment by comparing it with those achieved on a HPC in-house cluster. The results we present were achieved by using the Amazon Elastic Compute Cloud (EC2) of the Amazon Web Services (AWS) to process SAR datasets collected by the ENVISAT satellite and show that, thanks to the Cloud resources availability and flexibility, large DInSAR data volumes can be processed through the P-SBAS algorithm in short time frames and at reduced costs. As a case study, the mean deformation velocity map of the southern California area has been generated by processing 172 ENVISAT images. By exploiting 32 EC2 instances this processing took less than 17 hours to complete, with a cost of USD 850. Considering the available PB-scale archives of SAR data and the upcoming huge SAR data flow relevant to the recently launched (April 2014) Sentinel-1A and the forthcoming Sentinel-1B satellites, the exploitation of Cloud Computing solutions is particularly relevant because of the possibility to provide Cloud-based multi-user services allowing worldwide scientists to quickly process SAR data and to manage and access the achieved DInSAR results.

Research centers performing biomedical research collide with the problem of the treatment of large amounts of data. Several scientific fields in biomedical adopt technologies that can analyze samples in a more accurate way thanks to the high granularity which the current equipment provide the results. The cloud computing technology allows to create scalable and flexible infrastructures and data management services. In recent years the number of solutions that can be included within the phenomenon of cloud computing has increased. There are many cases of distributed solutions with high storage and processing capacity and the possibility of serving a large number of users. In this paper authors describe a biological networks modeling tool, implemented with the aid of MapReduce algorithms that works on a cluster in a cloud computing infrastructure.

Cloud technologies are still characterized by critical issues, which pose specific challenges for application developers and operators. In particular cloud application-level and infrastructure-level are completely decoupled both in the development and runtime phases leading in poor QoS cloud services. Main issues related to the optimize the use of the hardware resources are partially solve with virtualization technologies but innovative methodology in the automatic management of resources, applications provisioning and deployment are urgently needed. This paper presents ALM Automation Platform over CHEF framework in context of services virtualization for Public Administration where typically a large number of technologies heterogeneous resources, applications are deployed and managed.

IDIPOS, that stands for Italian Database Infrastructure for Polar Observation Sciences, has been conceived to realize a feasibility study on infrastructure devoted to management of data coming from Polar areas. This framework adopted a modular approach identifying two main parts: the first one defines main components of infrastructure, and, the latter selects possible cloud solutions to manage and organize these components. The main purpose is the creation of a scalable and flexible infrastructure for the exchange of scientific data from various application fields. The envisaged infrastructure is based on the cutting-edge technology of the Community Cloud Infrastructure for an aggregation and federation of resources, to optimize the use of hardware. The infrastructure is composed of: a central node, several nodes distributed in Italy, interconnection between other systems realized in Polar areas. This paper aims to investigate cloud solution, and explore the key factors which may influence cloud adoption in the project such as scalability, flexibility and expandability. In particular, main cloud aspects addressed are related to data storage, data management, data analysis, infrastructure federation following recommendations from the Cloud Expert Group to allow sharing information in scientific communities.

Often scientific applications are characterized by complex workflows and large datasets to manage. Usually, these applications run in dedicated high performance computing centers with low-latency interconnections which require a consistent initial cost. Public and private cloud computing environments, thanks to their features such as customized computing environments, flexibility, and elasticity represent a valid alternative with respect to HPC clusters in order to minimize costs and optimize processing. In this paper the migration of an advanced Differential Synthetic Aperture Radar Interferometry (DInSAR) methodology for the investigation of Earth surface deformation phenomena to the Amazon Web Services (AWS) cloud computing environment is presented. Such a technique which is referred to as Parallel Small Baseline Subset (P-SBAS) algorithm allows producing mean deformation velocity maps and the corresponding displacement time-series from a temporal sequence of radar images. Moreover, an experimental analysis aimed at evaluating the P-SBAS algorithm parallel performances which are achieved within the AWS cloud by exploiting two different families of instances and by taking into account different I/O and network bandwidth configurations is presented.

Cloud computing is a convenient, on demand, model which relies on shared pool of computing resources that can be rapidly provisioned and needs minimal management effort. As cloud computing obtains popularity nowadays, customers are looking for cloud solutions to adapt their organization's requirements. The more changes occur, customers are trying to decide what kind of cloud they are going to choose. Beside that, users are concerned about security, privacy, vendor lock-in and cost issues. A hybrid cloud is a mix of one private cloud and at least one public cloud, customers could benefit interoperability and flexibility. There are many elements that should be considered in hybrid cloud architecture. This paper represents a set of mpi benchmarking test executed on a private and public cloud environment in order to evaluate the QoS when real application is running.

As cloud computing adoption and deployment increase, the performance evaluation of the cloud environments is becoming very important. Cloud applications have different composition, configuration, and deployment requirements. Simulation and modeling techniques are suitable to quantify the performance of resource allocation policies and application scheduling algorithms in Cloud computing environments for different application and service models according to different work loads, energy performance and system size. In this paper, we give an overview of the existing distributed systems simulation and modeling tools in order to outline the main characteristics and peculiarities. We then present an outlook on new requirements to be addressed for performance evaluation of cloud applications through simulation and modeling.

The advanced Differential SAR Interferometers (DInSAR) methodologies are widely used for the investigation of Earth's surface deformation phenomena. In particular, the advanced DInSAR approach referred to as Small Baseline Subset (SBAS) technique is able to produce deformation velocity maps and the corresponding displacement time-series from a temporal sequence of space borne SAR acquisitions. Considering the already huge SAR data archives as well the upcoming massive data flow coming from the SENTINEL satellite constellation, cloud computing can be a valid solution to carry out DInSAR analyses thanks to its scalability and flexibility features. In this paper, the focus is given on the migration of the whole parallel version of the SBAS technique, namely P-SBAS, to a cloud environment by taking into account different parameters that influence processing time. Experimental tests that have been performed using both private and public cloud are also presented.

The ionosphere is the single largest contributor to the GNSS (Global Navigation Satellite System) error budget and ionospheric scintillation (IS) in particular is one of its most harmful effects. The Ground Based Scintillation Climatology (GBSC) has been recently developed by INGV as a software tool to identify the main areas of the ionosphere in which IS is more likely to occur. Due to the high computational load required, GBSC is currently used only for scientific, offline, studies and not as a real time service. Recently, a collaboration was initiated between ISMB and INGV in order to identify which cloud service model (IaaS, PaaS or SaaS) is most suitable for implementing the GBSC technique within the cloud computing environment. The aims of this joined effort are twofold: i) to optimize the computational resources allocation strategy/plan for the GBSC service, ii) to fine tune the algorithm for dynamic and real time application, towards a service contributing to high precision professional applications for the GNSS-reliant business sectors. Preliminary result of the implementation of GBSC within the cloud environment will be shown.

Data as a Service (DaaS) is among the latest kind of services being investigated in the Cloud computing community. The main aim of DaaS is to overcome limitations of state-of-the-art approaches in data technologies, according to which data is stored and accessed from repositories whose location is known and is relevant for sharing and processing. Besides limitations for the data sharing, current approaches also do not achieve to fully separate/decouple software services from data and thus impose limitations in inter-operability. In this paper we propose a DaaS approach for intelligent sharing and processing of large data collections with the aim of abstracting the data location (by making it relevant to the needs of sharing and accessing) and to fully decouple the data and its processing. The aim of our approach is to build a Cloud computing platform, offering DaaS to support large communities of users that need to share, access, and process the data for collectively building knowledge from data. We exemplify the approach from large data collections from health and biology domains.

Massive Parallel Sequencing is a term used to describe several revolutionary approaches to DNA sequencing, the so-called Next Generation Sequencing technologies. These technologies generate millions of short sequence fragments in a single run and can be used to measure levels of gene expression and to identify novel splice variants of genes allowing more accurate analysis. The proposed solution provides novelty on two fields, firstly an optimization of the read mapping algorithm has been designed, in order to parallelize processes, secondly an implementation of an architecture that consists of a Grid platform, composed of physical nodes, a Virtual platform, composed of virtual nodes set up on demand, and a scheduler that allows to integrate the two platforms.

Cloud computing is becoming increasingly popular for the provisioning of computing resources, in particular, through scientific tools that perform modeling or simulations. Vehicular Ad-Hoc Networks (VANETs) are mobile ad hoc networks which are meant to support primarily safety warnings and to manage challenging conditions to improve our transportation experience. This is a challenging context where large amounts of data need to be elaborated and analyzed in order to fully understand protocol and phenomena behaviors. In fact, VANET simulations are typically computationally intensive problems, and lend themselves for execution on distributed systems. A new hybrid cloud infrastructure is here presented to help and support the simulation science. This architecture optimizes the scheduling and execution of a batch of simulations, increasing the overall performance, in terms of simulation time and costs. Results clearly highlight the potentiality of this technology, proving as a valuable tool for network simulations.

Vehicular Ad-hoc NETworks (VANETs) are ad hoc networks aimed at improving the safety and efficiency of transportation in the near future. Despite the availability of results from field trials, simulations still play an unequalled role in the comprehensive understanding of complex and crowded VANET scenarios. Even more, VANET simulations are typically computationally intensive problems and lend themselves for execution on distributed systems. This paper presents a new architecture optimizing the scheduling and execution of a batch of simulations over a hybrid cloud. Results reveal that, in case of multiple simulations to be executed, the overall performance can deeply benefit from a distributed approach, reducing time and costs.

Distributed architecture is a fundamental way for science and engineering data application by aggregate power of computing resources connected by networks of the grid. This paper describes a innovative scenarios for intensive rainfall data analysis. The Flood Early Warning System is performed to give an alert in advance about the occurrence of heavy rainfalls around the world. The aim of the project is to evaluate the critical flood events and understand the potentially floodable areas where their assistance is needed. The World Food Programme and other humanitarian assistance organizations needs to understand the potential alert in order to organize the logistic helps. This system is based on precipitation analysis and it uses rainfall data from satellite at worldwide extent. The analysis and alert are delivered in near real-time to monitor the current rainfall condition over the world. Considering the global coverage area and the great deal of data to process, this paper presents a infrastructure solution based on grid computing technology. Our focus is on the advantages of using a distributed architecture in terms of performances in this specific context.

GPS - Radio Occultation is a remote sensing technique aim to characterize some atmospheric parameters. It is based on the analysis of the time evolution measurable on the GPS signal received on-board a Low Earth Orbit satellite. Result of analysis is the atmospheric refractivity profile which, in turn, can be used to characterize temperature, pressure and humidity profiles. This paper present the ROSA-ROSSA software (ROSA-Research and Operational Satellite and Software Activities) and a new approach for the Radio Occultation data processing by using Grid Computing technique.