1,721,203 research outputs found
DIFFUSE: A DIstributed and decentralized platForm enabling Function composition in Serverless Environments
Full text linkServerless computing is an emerging proposition in the cloud offering landscape that promotes a higher level of abstraction, further decoupling software operations from the underlying hardware. Often recognized as an economically driven computational approach, the serverless model relies on the execution of short-lived stateless functions, enabling a fine-grained accounting and control of resources. In this context, function composition represents an appealing feature, allowing the composition of two or more functions to create tailored processing pipelines, incentivizing modularity and reusability of functions, while paving the way to application-specific run-time optimizations. This work presents DIFFUSE: a DIstributed and decentralized platForm enabling Function composition in Serverless Environments. DIFFUSE embodies an innovative infrastructural support, enabling the efficient and transparent composition of functions by relying on pluggable middleware support, serving as a conveyor of messages among the platform components. Broadening the deployment spectrum of our proposal, we assess different middleware solutions, each presenting distinct delivery profiles, evidencing the tradeoffs that emerge
Context data distribution in mobile systems: A case study on Android-based phones
No full textContext awareness, namely the provisioning of the current execution context at the application level, forces the continuous delivery of context data to resource-constrained mobile devices, and that can become too severe a constraint even for modern support (Android, iOS, etc.). This article focuses on the realization of a context data distribution infrastructure for Android-based mobile phones, and highlights important details on the implementation of specific context distribution primitives. Finally, we present new experimental results to assess the runtime performances obtainable with a real Android deployment
Smart Management of Healthcare Professionals Involved in COVID-19 Contrast With SWAPS
Full text linkThe recent COVID-19 pandemic in Italy has highlighted several critical issues in the management process of infected people. At the health level, the management of the COVID-19 positive was mainly delegated to the regional authorities and centrally monitored by the State. Despite requested common activities (such as diagnosis of virus positivity, active surveillance of infected people and contact tracing), Regional Health Departments were able to issue specific directives in their territories and establish priority levels for each activity according to the specific needs related to the emergency in their area. The development of novel digital tools for the management of infected people become an urgent necessity to foster more organized and integrated solutions, able to quickly process large amounts of data. Mobile Crowdsensing methodologies could effectively facilitate needed lateral interviewing activities as well as the monitoring of crowds in environments with a high concentration of virus-positive subjects (such are hospital wards but also other locations), facilitating the tracing of possible outbreaks of contagion due to advanced geolocation techniques and big data analysis methods. This paper analyzes the functionality of SWAPS (Supporting Workflows for Healthcare Personnel management), a modular and scalable web platform which facilitate and reduces the management time of COVID positive health personnel within healthcare facilities. It also analyzes the possible integrations between SWAPS and ParticipACT, an advanced MCS platform developed by the University of Bologna that can help set up the alert notification in case of entry into a COVID risk area. This article surveys the current literature on software platforms to address COVID-19 and related tracing issues and presents the practical issues and on-the-field results obtained from the research developed by the University of Bologna by assisting the deployment of the proposed solution for a big Regional Health Department in the city of Bologna
A Support Infrastructure for Machine Learning at the Edge in Smart City Surveillance
No full textNowadays, the massive usage of mobile and IoT applications generate large amounts of data. Due to several reasons, including latency and bandwidth, it is not practical to send all generated data to the cloud. Recent standardization efforts, namely, Fog computing and the Multi-access Edge Computing (MEC), provide an extension of Cloud computing storage and network resources placed in a geographically distributed manner at the edge of the network closer to mobiles and IoT devices. These paradigms allow low latency, high bandwidth, and location-based awareness. In this paper, we present an infrastructure to support distributed Machine Learning (ML) by enabling edge devices to collaboratively learn a shared model while keeping local knowledge stored at the edge of the network. In addition, we claim the possibility of improving the model through the cloud that acts as a supervisor of the system that contains the global knowledge of the entire system through the integration of local edge models. We describe our architectural proposal and analyze a case study, namely video streaming processing for face recognition, deployed in a collaborative edge network. Finally, we report experimental results that show the potential advantages of using our approach instead of ML algorithms completely expected at the cloud infrastructure
Balanced Partitions of Trees and Applications
Full text linkWe study the k-BALANCED PARTITIONING problem in which the vertices of a graph are to be partitioned into k sets of size at most ceil(n/k) while minimising the cut size, which is the number of edges connecting vertices in different sets. The problem is well studied for general graphs, for which it cannot be approximated within any factor in polynomial time. However, little is known about restricted graph classes. We show that for trees k-BALANCED PARTITIONING remains surprisingly hard. In particular, approximating the cut size is APX-hard even if the maximum degree of the tree is constant. If instead the diameter of the tree is bounded by a constant, we show
that it is NP-hard to approximate the cut size within n^c, for any constant c<1.
In the face of the hardness results, we show that allowing near-balanced solutions, in which there are at most (1+eps)ceil(n/k)
vertices in any of the k sets, admits a PTAS for trees. Remarkably, the computed cut size is no larger than that of an optimal balanced solution. In the final section of our paper, we harness results on embedding graph metrics into tree metrics to extend our PTAS for trees to general graphs. In addition to being conceptually simpler and easier to analyse, our scheme improves the best factor known on the cut size of near-balanced solutions from O(log^{1.5}(n)/eps^2) [Andreev and Räcke TCS 2006] to 0(log n), for weighted graphs. This also settles a question posed by Andreev and Räcke of whether an algorithm with approximation guarantees on the cut size independent from eps exists
Data Distribution Service (DDS): A performance comparison of OpenSplice and RTI implementations
No full textData distributions systems with guaranteed Quality of Service (QoS) levels, such as the data-centric Data Distribution Service (DDS) standard specification, have gained more and more success in the last decade. These systems represent suitable solutions for effective and high-performance data communication for challenging application scenarios with real-time requirements, such as air traffic management, industrial automation, smart grids, and, more recently, financial applications. Notwithstanding the last decade has witnessed the diffusion and consolidation of some major implementations, only a very few, in some sense obsolete, performance analysis studies are available in the literature. To fill that gap and to facilitate future IT decision processes, we propose a thorough analysis of the DDS implementations proposed by the two main stakeholders in the DDS market, namely, PrismTech and Real-Time Innovations (RTI). The reported experimental results point out the pros and cons of both solutions in terms of data delivery performance, also by precisely evaluating bottlenecks and overhead, for instance in terms of CPU and memory resource usage
Management Infrastructures for Power-Efficient Cloud Computing Architectures
No full textThe surging demand for inexpensive and scalable IT infrastructures has led to the widespread adoption of Cloud computing architectures. These architectures have therefore reached their momentum due to inherent capacity of simplification in IT infrastructure building and maintenance, by making related costs easily accountable and paid on a pay-per-use basis. Cloud providers strive to host as many service providers as possible to increase their economical income and, toward that goal, exploit virtualization techniques to enable the provisioning of multiple virtual machines (VMs), possibly belonging to different service providers, on the same host. At the same time, virtualization technologies enable runtime VM migration that is very useful to dynamically manage Cloud resources. Leveraging these features, data center management infrastructures can allocate running VMs on as few hosts as possible, so to reduce total power consumption by switching off not required servers. This chapter presents and discusses management infrastructures for power-efficient Cloud architectures. Power efficiency relates to the amount of power required to run a particular workload on the Cloud and pushes toward greedy consolidation of VMs. However, because Cloud providers offer Service-Level Agreements (SLAs) that need to be enforced to prevent unacceptable runtime performance, the design and the implementation of a management infrastructure for power-efficient Cloud architectures are extremely complex tasks and have to deal with heterogeneous aspects, e.g., SLA representation and enforcement, runtime reconfigurations, and workload prediction. This chapter aims at presenting the current state of the art of powerefficient management infrastructure for Cloud, by carefully considering main realization issues, design guidelines, and design choices. In addition, after an in-depth presentation of related works in this area, it presents some novel experimental results to better stress the complexities introduced by power-efficient management infrastructure for Cloud
Automated Provisioning of SaaS Applications over IaaS-Based Cloud SystemsAdvances in Service-Oriented and Cloud Computing
No full textSoftware as a Service (SaaS) applications fully exploit the potential of elastic Cloud computing Infrastructure as a Service (IaaS) platforms by enabling new highly dynamic Cloud provisioning scenarios where application providers could decide to change the placement of IT service components at runtime, such as moving computational resources close to storage so to improve SaaS responsiveness. These highly dynamic scenarios require automating the whole SaaS provisioning cycle spanning from resource management to dynamic IT service components placement, and from software deployment to enable needed component re-activation and rebinding operations. However, notwithstanding the core importance of these functions to truly enable the deployment of complex SaaS over IaaS environments, at the current stage only partial and ad-hoc solutions are available. This paper presents a support infrastructure aimed to facilitate the composition of heterogeneous resources, such as single Virtual Machines (VMs), DB services and storage, and stand-alone services, by automating the provisioning of complex SaaS applications over the widely diffused real-world open-source OpenStack IaaS
- …
