Search

Containers are a form of software virtualization, rapidly becoming the de facto way of providing edge computing services. Research on container-based edge computing is plentiful, and this has been buoyed by the increasing demand for single digit, milliseconds latency computations. A container scheduler is part of the architecture that is used to manage and orchestrate multiple container-based applications on heterogenous computing nodes. The scheduler decides how incoming computing requests are allocated to containers, which edge nodes the containers are placed on, and where already deployed containers are migrated to. This paper aims to clarify the concept of container placement and migration in edge servers and the scheduling models that have been developed for this purpose. The study illuminates the frameworks and algorithms upon which the scheduling models are built. To convert the problem to one that can be solved using an algorithm, the container placement problem in mostly abstracted using multi-objective optimization models or graph network models. The scheduling algorithms are predominantly heuristic-based algorithms, which are able to arrive at sub-optimal solutions very quickly. There is paucity of container scheduling models that consider distributed edge computing tasks. Research in decentralized scheduling systems is gaining momentum and the future outlook is in scheduling containers for mobile edge nodes.

This paper proposes an energy management system (EMS) architecture based on the Kubernetes container cluster to solve the problem of traditional EMSs being unable to simultaneously achieve high reliability and high resource utilization. Container cluster technology is used to encapsulate, isolate and deploy applications, which solves the problem of low system reliability caused by interlocking failures. Discrete Markov theory is applied to propose a dynamic Pod fault-tolerant EMS model. The results of the solution model are used to adjust the Pod redundancy in real-time to achieve the highest reliability to satisfy physical resource constraints. The results of the performance analysis show that the reliability of the proposed architecture is 99.9999504%. Compared with the EMS of the service-oriented architecture (SOA), the annual failure time is reduced from 3.83 minutes to 0.26 minutes. The comprehensive utilization of hardware resources increases by approximately 20%, and performance indicators such as the peak access success rate improve significantly. The proposed architecture is implemented in a real-power system, with good operating results and broad application prospects.

Decision-makers in synchromodal transport (ST) have different preferences toward different objectives, such as cost, time, and emissions. To solve the conflicts among objectives and obtain preferred solutions, a preference-based multi-objective optimization model is developed. In ST, containers need to be transferred across modes, therefore the optimization problem is formulated as a pickup and delivery problem with transshipment. The preferences of decision-makers are usually expressed in linguistic terms, so weight intervals, that is, minimum and maximum weights, are assigned to objectives to represent such vague preferences. An adaptive large neighborhood search is developed and used to obtain non-dominated solutions to construct the Pareto frontier. Moreover, synchronization is an important feature of ST and it makes available resources fully utilized. Therefore, four synchronization cases are identified and studied to make outgoing vehicles cooperate with changes of incoming vehicles’ schedules at transshipment terminals. Case studies in the Rhine-Alpine corridor are designed and the results show that the proposed approach provides non-dominated solutions which are in line with preferences. Moreover, the mode share under different preferences is analyzed, which signals that different sustainability policies in transportation will influence the mode share.

Nowadays, the usage of the fog cloud-based Internet of Things (IoT) applications among users has been growing progressively. These applications may be E-Transport, E-Healthcare, Augmented Reality, and 3D-Game. Generally, contemporary cloud frameworks offer services based on virtual machines. These frameworks incurred with following issues such as long boot-time, overhead and unnecessary cost to run the IoT applications. We propose a new Microservice container fog system (MSCFS) based framework to run the mobility and delay-sensitive applications with minimum cost. The study the cost-efficient task scheduling problem in the heterogeneous fog servers. Furthermore, the study introduces the Cost Aware Computational Offloading and task scheduling (CACOTS)framework, which solves the task scheduling into multiple steps. Such as task sequencing step, resource matching step and scheduling step. The experimental results prove that the proposed MSCFS and CACOTS schemes can enhance server utilization. As decrease the services latency and average services bootup time more effectively, and minimize costs.

Autonomous driving is one of the most innovative applications nowadays. However, autonomous driving is still suffering from heavy calculation, high energy consumption and strict real-time execution constraints. Different from cloud computing, edge computing deploys calculation, storage and service on the edge of network. It is a better platform to serve efficiency and privacy oriented autonomous driving service offloading. To this end, we proposed a container-based edge offloading framework for autonomous driving. This framework builds an Offloading Decision Module, an Offloading Scheduler Module and an Edge Offloading Middleware on top of the lightweight virtualization. It provides the abstraction and management of the execution environment in the granularity of containers on edge. Therefore, it enables the privacy preserve and resource isolation for autonomous driving execution constraints. Its utility preferable offloading schedule strategy formalized the multi-application multi-edge nodes mapping problem into a multiple multidimensional knapsack problem (MMKP) and gave a utility oriented greedy algorithm (GA) for real-time solving. The experimental results show that the proposed framework has high feasibility and isolation meanwhile can guarantee millisecond-level autonomous driving offloading on edge.

Since the development of Docker in 2013, container utilization projects have emerged in various fields. Docker has the advantage of being able to quickly share application build environments among developers through container technology, but it does not provide security guarantees for known security vulnerabilities inside Docker images. Since the Docker images are shared without a means of security vulnerability diagnostic, polluted Docker images can be distributed so that the Docker-based application build environments can be easily collapsed. In this paper, we introduce a Docker Image Vulnerability Diagnostic System (DIVDS) for a reliable Docker environment. The proposed DIVDS diagnoses Docker images when uploading or downloading the Docker images from a Docker image repository.

Container shipping plays an important role in economic development by providing services for unitized cargos on fixed ocean trading routes between destinations. The transit time, as well as freight cost, presents temporal and monetary costs in the global trade system. Given the fixed routing characteristic in container shipping, a study on the reliability of travel times is crucial to understand the dynamics on temporal cost in major international trades. This research investigates empirically whether there exist spatial differences in international container shipping routes by employing actual shipment data for a fleet of over 600 container vessels extracted from the Automatic Identification System (AIS). The results show that the geographic distribution of the routes have significant impact on travel time uncertainty and hence there is clear variation across international trade routes. Sailing route security, infrastructure and trade volumes in the routes are likely the underlying explanations. This research contributes to the literature in two folds. We pioneer in modelling the micro-level container invisible trade cost by computing the service uncertainty. This also provides insight of the temporal cost of international container cargo trade in terms timing reliability and its dynamic structures. Published version This research is funded by the Research Council of Norway under the project ‘Real Energy Efficiency and Emissions in the Seaway’, project no. 255672/O80

The right to safety of purchased goods is one of the most important consumer rights. Therefore, an accurate analysis of safety of machinery and equipment purchased on the secondary market is an extremely important issue. One of its elements should be the assessment of the degree of service live usage. In the case of cranes, the subject of such an assessment covers both the supporting structure of the device and its mechanisms and installations. Information on evaluation of the service life expiration of the secondhand cranes may be necessary and support the decision-making process associated with the purchase of the device. The paper deals with the problem of evaluating the expiration of the service life of the supporting structure of cranes coming from the secondhand market. The result of such evaluation makes it possible to determine whether the production capacity of a device has been exhausted. This forms the basis for the decisions on any possible further operation of the crane. The work presents an approach based on the provisions of standard ISO 12482: 2014. This document is the basis for service life evaluation for cranes. The approach presented in it makes the way of analysis dependent on the method used to collect operational data. In this paper the evaluation of service life expiration was prepared on the example of a ship to shore gantry cranes. These devices are characterized by particularly large dimensions and weight. Damage of their supporting structure can lead to the significant losses. Some of these devices change the location during operation. Therefore, it is important to know the current level of their service life usage. A case, in which the operational history of a device is not known, was considered. In the paper the special attention was paid to the methodology of estimating the load distribution coefficient. In order to describe the hypothetical load spectrum, triangular distribution was proposed. The results of the calculations for the structure of port cranes are presented and discussed.

International audience; One of the cornerstones of the cloud provider business is to reduce hardware resources cost by maximizing their utilization. This is done through smartly sharing processor, memory, network and storage, while fully satisfying SLOs negotiated with customers. For the storage part, while SSDs are increasingly deployed in data centers mainly for their performance and energy efficiency, their internal mechanisms may cause a dramatic SLO violation. In effect, we measured that I/O interference may induce a 10x performance drop. We are building a framework based on autonomic computing which aims to achieve intelligent container placement on storage systems by preventing bad I/O interference scenarios. One prerequisite to such a framework is to design SSD performance models that take into account interactions between running processes/containers, the operating system and the SSD. These interactions are complex. In this paper, we investigate the use of machine learning for building such models in a container based Cloud environment. We have investigated five popular machine learning algorithms along with six different I/O intensive applications and benchmarks. We analyzed the prediction accuracy, the learning curve, the feature importance and the training time of the tested algorithms on four different SSD models. Beyond describing modeling component of our framework, this paper aims to provide insights for cloud providers to implement SLO compliant container placement algorithms on SSDs. Our machine learning-based framework succeeded in modeling I/O interference with a median Normalized Root-Mean-Square Error (NRMSE) of 2.5%.

The aftermath of the Indian Ocean tsunami on 26 December 2004 triggered by the off Sumatra earthquake (magnitude “M” = 9.1), and the Great East Japan earthquake of 11 March 2011 off the Pacific coast of Tohoku (M = 9.0), evidence the secondary damage from driftage collision due to large tsunami waves. To prevent this type of damage, the establishment of methods for predicting driftage movement and calculating the impact force by driftage is necessary. Several numerical models have been developed to predict the driftage movement of objects. Every year, these improve in accuracy and usability. In contrast, there are many calculation formulae for calculating the impact force. However, since there are considerable differences between values calculated using these formulae, the reliability of each formula is unknown. Therefore, in this research, one team of the committee on tsunami research of the Japan Society of Civil Engineers summarizes the main calculation formulae of impact forces that have been proposed until 2019. In addition, for each type of driftage (driftwood, containers, cars, ships), we compare calculation values of these formulae with measured data of large-scale experiments. Finally, we check the range of calculation values for each formula up to 15 m/s in collision velocity and clarify then the following facts: (1) In the case of driftwood, the formulae of Matsutomi, Federal Emergency Management Agency (FEMA) and National Oceanic and Atmospheric Administration (NOAA), and American Society of Civil Engineers (ASCE) are most reliable, (2) In the case of containers, the formulae of Matsutomi, Arikawa et al., FEMA and NOAA, Ikeno et al., and ASCE are most reliable, (3) In the case of cars, the formulae of FEMA and NOAA, and ASCE are most reliable, (4) In the case of ships, the formulae of Mizutani, FEMA and NOAA, and ASCE are most reliable.

Currently, cloud service providers have used a variety of operational mechanisms to support the company's business processes. Therefore, the services are stored on the company's server, which presents in the form of infrastructure, platform, software, and function. There are several vulnerabilities that have been faced in the implementation such as system failure, natural disasters, human errors, or attacks from unauthorized parties. In addition, the time of unavailability of services can be minimized by doing an LM, in which many servers have been used the containers to behave like a service provider. Actually, its existence replaces the virtual machine that requires more resources although the process only can be done through docker with checkpoint and restore in userspace (CRIU). In this research, LM processes are done to the docker container using CRIU by analyzing the quality of service (QoS), memory, and CPU usage. Thus, the simulation is carried out by establishing the LM using 2 (two) different platforms through scenarios with one and three containers respectively. The performance analysis results aim to examine several indicators in comparison with the achievable result to reduce problems that occurred in the cloud service.

International audience; Compared to full virtualization, containerization reduces virtualization overhead and resource usage, offers reduced deployment latency and improves reusability. For these reasons, containerization is massively used in an increasing number of applications. However, because containers share a full kernel with the host, they are more vulnerable to attacks that may compromise the host and the other containers on the system.In this paper, we present SNAPPY (Safe Namespaceable And Programmable PolicY), a new framework that allows even unprivileged processes such as containers to safely and dynamically enforce in the kernel fine-grained, stackable and programmable eBPF security policies at runtime. This is done by making working coordinately a new LSM (Linux Security Module) Module, a new security Linux namespace abstraction (policy_NS) and eBPF policies enriched with ’dynamic helpers’. This design especially allows to minimize containers’ attack surface. Our design may be applied to any processes but is particularly suitable for container-based use cases. We show that SNAPPY can effectively increase the security level of containers for different use cases, can be easily integrated with the most relevant norms (OCI, Open Container Initiative) and containerization engines (Docker and runC) and has a performance overhead lower than 0.09% in realistic scenarios.

This paper proposes Hermes, a container-based preemptive GPU scheduling framework for accelerating hyper-parameter optimization in deep learning (DL) clusters. Hermes accelerates hyper-parameter optimization by time-sharing between DL jobs and prioritizing jobs with more promising hyper-parameter combinations. Hermes’s scheduling policy is grounded on the observation that good hyper-parameter combinations converge quickly in the early phases of training. By giving higher priority to fast-converging containers, Hermes’s GPU preemption mechanism can accelerate training. This enables users to find optimal hyper-parameters faster without losing the progress of a container. We have implemented Hermes over Kubernetes and compared its performance against existing scheduling frameworks. Experiments show that Hermes reduces the time for hyper-parameter optimization up to 4.04 times against previously proposed scheduling policies such as FIFO, round-robin (RR), and SLAQ, with minimal time-sharing overhead.

Due to the growth of users and linked devices in networks, there is an emerging need for dynamic solutions to control and manage computing and network resources. This document proposes a Distributed Wireless Operative System on a Mobile Ad-hoc Network (MANET) to manage and control computing resources in relation to several virtual resources linked in a wireless network. This prototype has two elements: a local agent that works on each physical node to manage the computing resources (e.g., virtual resources and distributed applications) and an orchestrator agent that monitors, manages, and deploys policies on each physical node. These elements arrange the local and global computing resources to provide a quality service to the users of the Ad-hoc cluster. The proposed S.O.V.O.R.A. model (Operating Virtualized System oriented to Ad-hoc networks) defines primitives, commands, virtual structures, and modules to operate as a distributed wireless operating system.

International audience; The container mechanism amortizes costs by consolidating several servers onto the same machine, while keeping them mutually isolated.Specifically, to ensure performance isolation, Linux relies on memory limits.These limits are static, despite the fact that application needs are dynamic; this results in poor performance.To solve this issue, MemOpLight uses dynamic application feedback to rebalance physical memory allocation between containers focusing on under-performing ones.This paper presents the issues, explains the design of MemOpLight, and validates it experimentally.Our approach increases total satisfaction by 13% compared to the default.

The past few years have witnessed many software companies refactoring their monolithic applications into microservices to take advantage of containers in Cloud-native and DevOps-driven environments. Modern Cloud environments achieve high levels of scalability, flexibility, availability and business agility by combining two different kinds of virtualization: virtual machines and containers. The deployment of complex applications in such environments should be done by taking into account the interactions that may arise from the overlapping of these two different levels of virtualization. In this paper, we first describe a preliminary evaluation of a containerized application that has been obtained by refactoring a monolithic application into microservices. The application was deployed in a private Cloud system that relies on Kubernetes, which is a cluster management software for Docker containers that have been deployed on top of OpenStack-managed KVM virtual machines. Subsequently, in order to systematically investigate the possible interactions that can be produced when multiple containers running different kinds of workloads are activated on top of virtual machines, we present and analyze the results obtained by running different benchmarks in a testbed setup. From the analysis of the benchmark results, we derive some guidelines that should be taken into account by container schedulers when container orchestration platforms, such as Kubernetes, are deployed on top of a virtualized IaaS layer.

The container-based cloud is used in various service infrastructures as it is lighter and more portable than a virtual machine (VM)-based infrastructure and is configurable in both bare-metal and VM environments. The Internet-of-Things (IoT) cloud-computing infrastructure is also evolving from a VM-based to a container-based infrastructure. In IoT clouds, the service availability of the cloud infrastructure is more important for mission-critical IoT services, such as real-time health monitoring, vehicle-to-vehicle (V2V) communication, and industrial IoT, than for general computing services. However, in the container environment that runs on a VM, the current fault detection method only considers the container’s infra, thus limiting the level of availability necessary for the performance of mission-critical IoT cloud services. Therefore, in a container environment running on a VM, fault detection and recovery methods that consider both the VM and container levels are necessary. In this study, we analyze the fault-detection architecture in a container environment and designed and implemented a Fast Fault Detection Manager (FFDM) architecture using OpenStack and Kubernetes for realizing fast fault detection. Through performance measurements, we verified that the FFDM can improve the fault detection time by more than three times over the existing method.

Cloud applications are exposed to workloads whose intensity can change unpredictably over time. Hence, the ability to quickly scale the amount of computing resources provisioned to applications is essential to minimize costs while providing reliable services. In this context,containers are deemed to be a promising technology to enable fast elasticity in resource allocation schemes.In this paper, we propose and experimentally test an efficient container-based cloud computing provisioning system. First, we address the container deployment problem and discuss how to manage container provisioning and scaling. Second, we devise are source management mechanism leveraging on both admission control and auto-scaling techniques. We propose to drive auto-scaling decisions through a Q-Learning algorithm, which is agnostic to the specific computing environment, and proceeds based only on the load of the physical processors assigned to a container. We evaluate our solution in two experimental setups,and show that it yields significant advantages when compared to popular container managers such as Kubernetes. TRUE pub

Nowadays the heightened awareness of the critical trend in resource depletion impels to improve the eco &minus, sustainability of any productive process. The research presented in this paper aims to quantify the environmental impact of the emerging productive process of edible flowers, focusing on two model species, i.e., Begonia x semperflorens &minus, cultorum hort and Viola cornuta L., and two types of product, i.e., flowering potted plants sold in plastic vases and packaged flowers ready to be consumed. The study was carried out in an Italian nursery located in Tuscany, interviewing the owners in order to complete the Life Cycle Inventory, assessing the value of the impact categories, and using the &ldquo, cradle to gate&rdquo, approach. The information about the production of flowering potted plants and packaged flowers were inserted in a database and elaborated by the appropriate software. The results of the Life Cycle Assessment (LCA) analysis referred to 1 g of fresh edible flowers and were expressed in four impact categories. Global Warming Potential (GWP) values ranged from 24.94 to 31.25 g CO2 eq/g flowers, Acidification Potential (AP) ranged from 8.169E &minus, 02 to 1.249E &minus, 01 g SO2 eq/g flowers, Eutrophication Potential (EP) ranged from 3.961E &minus, 02 to 5.284E &minus, 02 g PO43 &minus, eq/g flowers, and Photochemical Ozone Creation Potential (POCP) ranged from 8.998E &minus, 03 to 1.134E &minus, 02 g C2H4 eq/g flowers. Begonias showed lower emissions than violas in the GWP and POCP indexes, whereas violas showed lower values in the AP and EP impact categories. The most impactful phase was the propagation, accounting on average for 42% of the total emissions. Overall, the findings highlighted a higher environmental load for the production of both begonias and violas packaged flowers, especially if in small containers, rather than as potted plants, with an emission percentage increase from 8% to 17% among the impact categories.

Events, such as prolonged congestion in ports or unavailable ship routes in the maritime network, often initiate cascading congestions that block transportation and/or disrupt services over wide areas. Existing traffic flow analysis methods lack the ability to understand the cascading effects of delays in ship routes or how to reduce overall delays in greater maritime areas. Dependency risk graphs have been proposed as a tool for analyzing such cascading events using dependency chains. This paper proposes a risk-based interdependency analysis method capable to detect large-scale traffic congestions between interconnected ports and ship routes in the maritime network and provide solutions to improve flow. Presented dependency risk chains of ports along with graph theory help us analyze ship routes and detect ports that are affected most when other major ports are congested in the maritime network, detect the causes of bottlenecks, and provide valuable info in relieving delays across container ship routes. We apply the proposed method on historical container ship routing data provided by the MarineTraffic company that maintains a comprehensive maritime database worldwide for more than six million users monthly. This application-oriented, interdisciplinary effort culminated in a prototype tool is able to analyze the historical data for container ships in the entire global maritime network and detect congestion dependencies. The tool can be used to identify key shipping routes or ports that: 1) are prone to delays; 2) greatly affect the overall maritime network due to position, connections and risk of congestion; and/or 3) get affected the most by delays in previous route legs.

The authors of this article presented technologies of spreading coatings over structural details of automated guided vehicles used in the handling terminals. Technological process of painting with the use of powder paints was presented, as well as innovative methods of surface preparation with the use of BONDERITE M-NT 2011 technology and OXILAN 9807 preparation. Modern methods of surface preparation with production of silane conversion coatings were presented and analysed. The authors of this article selected a system of guaranteed power supply of the processing lines. ABCUS software was applied to check the time of battery backup of UPS units in selected N+1 system.

We introduce an open-source, scalable, and distributed architecture, called Megalos, that supports the implementation of virtual network scenarios consisting of virtual devices (VDs) where each VD may have several Layer 2 interfaces assigned to virtual LANs. We rely on Docker containers to realize vendor-independent VDs and we leverage Kubernetes for the management of the nodes of a distributed cluster. Our architecture does not require platform-specific configurations and supports a seamless interconnection between the virtual environment and the physical one. Also, it guarantees the segregation of each virtual LAN traffic from the traffic of other LANs, from the cluster traffic, and from Internet traffic. Further, a packet is only sent to the cluster node containing the recipient VD. We produce several example applications where we emulate large network scenarios, with thousands of VDs and LANs. Finally, we experimentally show the scalability potential of Megalos by measuring the overhead of the distributed environment and of its signaling protocols.

Purpose. The study is aimed at determination of the loading of containers in mixed trains when transporting by train ferries. Methodology. In order to achieve the objective the author studied accelerations (as components of the dynamic load) influencing the bearing structure of a 1CC standard container located on a flat car during transportation by train ferry. In order to determine accelerations influencing the bearing structure of a container, a mathematical model of the container’s movements under train ferry oscillations was made. The model considered angular displacements of a train ferry relative to the longitudinal axis (bank) as being the maximum load on the bearing structure, as well as changes in the container stability relative to the flat car frame. While working out the motion equations three diagrams of interrelation between the container and the flat car located on the train ferry deck were considered: 1) absence of displacements of the flat car and containers relative to the initial place under the train ferry oscillations; 2) with displacements of the flat car under the train ferry oscillations when containers are immovable relative to the flat car frame; 3) with displacements of the flat car relative to the deck and relative to the flat car frame. The differential equations of motion were solved in Mathсad taking into account their reduction to a normal Cauchy problem with a subsequent integration by the Runge–Kutta method. Findings. Refined values of accelerations influencing the bearing structure of containers located on the flat car while transporting by train ferry were obtained. Originality. The mathematical models of displacements of container bearing structures located on a flat car while transporting by train ferry were proposed. Practical value. The results of investigations can be applied for designing bearing structures of new generation containers, besides they may improve the efficiency of mixed transportation along the international transport corridors.

The UK's highly privatised port system means that, while many of the issues in the port governance literature relevant to port concessions do not arise here, the respective roles of harbour authorities and port operators continue to be questioned. The concern in the UK is whose role it should be to monitor the capacity and service quality of the port sector, including how to govern the ways in which the different classes of port stakeholder interact. This paper describes and discusses the UK port sector, the main ports and cargo types, the governance system and recent developments. Recent changes in national policy are reviewed and potential new developments in governance are considered, reflecting on how the UK case represents some key theoretical considerations regarding infrastructure governance within a modern political system favouring private ownership and operation of the transport sector.

Mist computing extends the Internet of Things computing infrastructures to the IoT devices at the edges of the networks. The dynamic characteristics of the IoT environments and resource limitations of the devices introduce challenges to the orchestration of the mist platform resources. In this paper, we present a hybrid resource discovery solution for mist, based on the IETF CoRE Resource Directories deployed as containerized Microservices to the supporting edge devices. This way the directory instances can be deployed on-demand as part of the edge platform, where each instance serves a mist network connected to the hosting edge device. This enables low latency resource queries at one-hop distance for the mist applications. At the edge layer, the directories form a distributed discovery infrastructure, connecting resources in disparate mist networks with each other and cloud and edge applications. A real-world prototype of such discovery infrastructure is implemented, based on low resource edge devices hosting the directory instances and low-power embedded devices as the mist resource servers and clients. The prototype is evaluated with latency and power consumption measurements, where the results show that discovery latency is as low as half a millisecond with a low power consumption.

Most of the relevant Big Data processing frameworks (e.g., Apache Hadoop, Apache Spark) only support JVM (Java Virtual Machine) languages by default. In order to support non-JVM languages, subprocesses are created and connected to the framework using system pipes. With this technique, the impossibility of managing the data at thread level arises together with an important loss in the performance. To address this problem we introduce Ignis, a new Big Data framework that benefits from an elegant way to create multi-language executors managed through an RPC system. As a consequence, the new system is able to execute natively applications implemented using non-JVM languages. In addition, Ignis allows users to combine in the same application the benefits of implementing each computational task in the best suited programming language without additional overhead. The system runs completely inside Docker containers, isolating the execution environment from the physical machine. A comparison with Apache Spark shows the advantages of our proposal in terms of performance and scalability This work has been supported by MICINN, Spain (RTI2018-093336-B-C21), Xunta de Galicia, Spain (ED431G/08 and ED431C-2018/19) and European Regional Development Fund (ERDF) SI

We introduce an ETSI NFV compliant, scalable, and distributed architecture, called Megalos, that supports the implementation of virtual network scenarios consisting of virtual devices (VNFs) where each VNF may have several L2 interfaces assigned to virtual LANs. We rely on Docker containers to realize VNFs and we leverage Kubernetes for the management of the nodes of a distributed cluster. Our architecture guarantees the segregation of each virtual LAN traffic from the traffic of other LANs, from the cluster traffic, and from Internet traffic. Also, a packet is only sent to the cluster node containing the recipient VNF. The allocation of the VNFs to the nodes of the cluster is performed by Megalos Scheduler, taking into account the network topology in order to reduce the traffic among nodes. We produce an example application where we emulate a large network scenario, with thousands of VNFs and LANs, on a small cluster of 50 nodes. Finally, we experimentally show the scalability potential of Megalos by measuring the overhead of the distributed environment and of its signaling protocols.

This paper has been presented at 2020 European Conference on Networks and Communications (EuCNC). The fifth generation (5G) mobile network has paved the way for innovations across vertical industries. The integration of distributed intelligent edge into the 5G orchestrated architecture brings the benefits of low-latency and automation. A successful example of this integration is exhibited by the 5G-DIVE project, which aims at proving the technical merits and business value proposition of vertical industries such as autonomous drone surveillance and navigation. In this paper, and as part of 5G-DIVE, we present an aerial disaster relief system, called EagleEYE, which utilizes edge computing and machine learning to detect emergency situations in real-time. EagleEYE reduces training time by devising an object fusion mechanism which enables reusing existing datasets. Furthermore, EagleEYE parallelizes the detection tasks to enable real-time response. Finally, EagleEYE is evaluated in a real-world testbed and the results show that EagleEYE can reduce the inference latency by 90% with a high detection accuracy of 87%. This work has been partially funded by the H2020 EU/TW joint action 5G-DIVE (Grant #859881).

Understanding the resource usage behaviors of the ever-increasing machine learning workloads are critical to cloud providers offering Machine Learning (ML) services. Capable of auto-scaling resources for customer workloads can significantly improve resource utilization, thus greatly reducing the cost. Here we leverage the AI4DL framework [1] to characterize workload and discover resource consumption phases. We advance the existing technology to an incremental phase discovery method that applies to more general types of ML workload for both training and inference. We use a time-window MultiLayer Perceptron (MLP) to predict phases in containers with different types of workload. Then, we propose a predictive vertical auto-scaling policy to resize the container dynamically according to phase predictions. We evaluate our predictive auto-scaling policies on 561 long-running containers with multiple types of ML workloads. The predictive policy can reduce up to 38% of allocated CPU compared to the default resource provisioning policies by developers. By comparing our predictive policies with commonly used reactive auto-scaling policies, we find that they can accurately predict sudden phase transitions (with an F1-score of 0.92) and significantly reduce the number of out-of-memory errors (350 vs. 20). Besides, we show that the predictive auto-scaling policy maintains the number of resizing operations close to the best reactive policies.

As Cloud Computing (CC) branched areas such as Multi-access Edge Computing (MEC) and Fog computing are still on growing research interest. The creation of new tools to improve quality and speed the experimentation in such areas is a general interest. In this article, we propose COPA, an experimenter-level container orchestration tool for networking testbeds. This tool provides a friendly interface for the experimenter test container orchestration algorithms which can start, stop, copy, and even migrate a container from one host to another. COPA also includes network/resources monitoring to feed the experimenter’s orchestration algorithm so that it can make decisions based on real-time environment information. Furthermore, the experimenter can automatize the experiment scenario setup and deployment by pre-configuring in COPA. This tool helps the experimenter in testing different scenarios and quickly changing experiment parameters. Considering these features, COPA aims to provide an experimentation architecture to deploy and test container orchestration algorithms. Furthermore, we provide a case study explaining how COPA can be a key tool in the MEC and Network Function Virtualization (NFV) experimentation environments. This tool was already deployed in Federated Union of Telecommunications Research Facilities for an EU-Brazil Open Laboratory (FUTEBOL) testbeds as part of the control framework and was well validated by the project reviewers and partners.

Fog computing platforms offer virtualized resources located in the vicinity of their end users. Their broad geographical distribution force them to split physical resources in large numbers of relatively weak machines. The limited available disk space per fog node however creates problems for Docker-based systems which locally cache a copy of every container image they execute: first, caches may fill very quickly, whereas standard Docker never automatically evicts any image from its cache. This motivates us to implement automatic cache replacement in Docker. Second, splitting cache space in multiple disjoint partitions negatively impacts the hit rates. This motivates us to propose allowing multiple co-located Docker servers to share their caches. Our trace-based evaluations show that the proposed design achieves significant cache hit improvements, leading to reductions of average container deployment times between 37% and 78% depending on the scenarios.

In this paper, we describe how the microservices paradigm can be used to design and implement distributed edge services for Internet of Things applications. As a case study, traditionally monolithic user mobility analysis service is developed, with distributed and extendable microservices, for the standardized ETSI MEC system reference architecture. In each of the edge system three tiers, microservices implement the service logic with components for movement trace analysis, movement prediction and visualization of the results. The distributed service is implemented with Docker containers and evaluated on real-world settings with low capacity edge servers and real user mobility data. The results show that the edge promise of low latency can be met in such as implementation. The integration of a software development technology with a standardized edge system provides solid background for further development.

Nursery, floriculture, and propagation production accounted for 79% ($13.3 Billion) of 2017 ornamental specialty crop production in the United States. Access to high quality water sources is increasingly limited for irrigating these economically significant crops. Given the production, environmental, and economic issues associated with the use of water&mdash, including recycled, reclaimed, surface, and ground water&mdash, it is critical to develop sustainable runoff, containment, and remediation technologies, and to identify alternative sources of water. To better understand current practices and future water-related needs as perceived by grower stakeholders, an online survey was distributed nationally and five in-depth round table discussion sessions were conducted at the Mid-Atlantic Nursery Trade Show, Gulf States Horticultural Expo, California Grown Show, AmericanHort&rsquo, s Cultivate, and the Farwest Show with a total of 36 individual industry participants. A team of research and extension specialists facilitated by a Specialty Crops Research Initiative Planning Grant (NIFA Project # 2011-51181-30633) analyzed and concisely summarized the results from the survey and the round table discussions. Research priorities related to water management identified by stakeholders revolved around six themes: (1) recycled water infrastructure and management, (2) contaminants, (3) plant health and water quality, (4) water treatment technologies, (5) competing and complementary water uses, (6) societal perception of agricultural water use.

The virtualization technology has a great potential to improve the manageability and scalability of industrial control systems, as it can host and consolidate computing resources very efficiently. There accordingly have been efforts to utilize the virtualization technology for industrial control systems, but the research for virtualization of traditional industrial real-time networks, such as Controller Area Network (CAN), has been done in a very limited scope. Those traditional fieldbuses have distinguished characteristics from well-studied Ethernet-based networks, thus, it is necessary to study how to support their inherent functions transparently and how to guarantee Quality-of-Service (QoS) in virtualized environments. In this paper, we suggest a lightweight CAN virtualization technology for virtual controllers to tackle both functionality and QoS issues. We particularly target the virtual controllers that are containerized with an operating-system(OS)-based virtualization technology. In the functionality aspect, our virtualization technology provides virtual CAN interfaces and virtual CAN buses at the device driver level. In the QoS perspective, we provide a hierarchical real-time scheduler and a simulator, which enable the adjustment of phase offsets of virtual controllers and tasks. The experiment results show that our CAN virtualization has lower overheads than an existing approach up to 20%. Moreover, we show that the worst-case end-to-end delay could be reduced up to 78.7% by adjusting the phase offsets of virtual controllers and tasks.

International audience; The container mechanism supports server consolidation ; to ensure memory performance isolation, Linux relies on static memory limits. However, this results in poor performance, because an application needs are dynamic. In this article we will show current problems with memory consolidation for containers in Linux.

International audience; The general purpose of the present paper is to study the feasibility of removing brominated flame retardants (BFR) polymers from mixed granular wastes using tribo-electrostatic separation. The study was carried out on mixtures consisting in equal quantities of high impact polystyrene (HIPS) particles either BFR-free or containing 40000 ppm of brome (4.7%). Three triboelectric charging devices were used: a vibrating tribocharger (a container made of stainless steel, polystyrene or polypropylene and fixed on a vibrating table); a rotating cylinder tribocharger made of polyvinyl chloride; a fluidized bed tribocharger, the walls of which were made of poly methyl methacrylate. The selective sorting of the granular mixture was performed using a roll type electrostatic separator. The results confirm that separating BFR HIPS from BFR-free HIPS is feasible. Best separation results were obtained for the mixture tribocharged in the fluidized bed tribocharger.

Containerization is a lightweight virtualization technique reducing virtualization overhead and deployment latency compared to full VM; its popularity is quickly increasing. However, due to kernel sharing, containers provide less isolation than full VM. Thus, a compromised container may break out of its isolated context and gain root access to the host server. This is a huge concern, especially in multi-tenant cloud environments where we can find running on a single server containers serving very different purposes, such as banking microservices, compute nodes or honeypots. Thus, containers with specific security needs should be able to tune their own security level. Because OS-level defense approaches inherited from time-sharing OS generally requires administrator rights and aim to protect the entire system, they are not fully suitable to protect usermode containers. Research recently made several contributions to deliver enhanced security to containers from host OS level to (partially) solve these challenges. In this survey, we propose a new taxonomy on container defense at the infrastructure level with a particular focus on the virtualization boundary, where interactions between kernel and containers take place. We then classify the most promising defense frameworks into these categories.

International audience; With the increasing popularity of containers, managing them on top of virtual machines becomes a common practice, called nested virtualization. This paper presents BrFusion and Hostlo, two solutions that address each of two networking issues of nested virtualization: network virtualization duplication and virtual machine-bounded pod deployments. The first issue lengthens network packet paths while the second issue leads to resource fragmentation. For instance, in respect with the first issue, we measured a throughput degradation of about 68% and a latency increase of about 31% in comparison with a single networking layer. We prototype BrFusion and Hostlo in Linux KVM/QEMU, Docker and Kubernetes systems. The evaluation results show that BrFusion leads to the same performance as a single-layer virtualization deployment. Concerning Hostlo, the results show that more than 11% of cloud clients see their cloud utilization cost reduced by down to 40%.

The increasing of connected tiny, low-power, embedded devices, grouped into the generic definition of "Internet of Things" (IoT), raised remarkably the amount of in-situ collected data. However, at the time of writing, IoT devices have limited storage and computation resources if compared with a cloud computing or on-premises infrastructure. IoT devices often suffer for reduced connectivity due to the place of the deployment or other technical, environmental or economic reasons. In this work, we present the DagOn* workflow engine as a part of an IoT orchestration scenario oriented to operational environmental prediction. Our novel approach is devoted to join the two worlds of workflows, in which each task runs on a dynamically allocated computational infrastructure, with tiny jobs targeted to embedded devices hosting sensors and actuators. We show our preliminary results applied to a demonstration use case. We are confident that further development of the proposed technology will affect positively production applications for massive and geographically distributed data collection.

Recently, “Big Data” platform technologies have become crucial for distributed processing of diverse unstructured or semi-structured data as the amount of data generated increases rapidly. In order to effectively manage these Big Data, Cloud Computing has been playing an important role by providing scalable data storage and computing resources for competitive and economical Big Data processing. Accordingly, server virtualization technologies that are the cornerstone of Cloud Computing have attracted a lot of research interests. However, conventional hypervisor-based virtualization can cause performance degradation problems due to its heavily loaded guest operating systems and rigid resource allocations. On the other hand, container-based virtualization technology can provide the same level of service faster with a lightweight capacity by effectively eliminating the guest OS layers. In addition, container-based virtualization enables efficient cloud resource management by dynamically adjusting the allocated computing resources (e.g., CPU and memory) during the runtime through “Vertical Elasticity”. In this paper, we present our practice and experience of employing an adaptive resource utilization scheme for Big Data workloads in container-based cloud environments by leveraging the vertical elasticity of Docker, a representative container-based virtualization technique. We perform extensive experiments running several Big Data workloads on representative Big Data platforms: Apache Hadoop and Spark. During the workload executions, our adaptive resource utilization scheme periodically monitors the resource usage patterns of running containers and dynamically adjusts allocated computing resources that could result in substantial improvements in the overall system throughput.

This work discusses the problem of dangerous containers storage in a container terminal. Container terminal represents an essential intermodal interfaces for global transportation network. Several materials handling possible to move containers at the port to better meet the needs of ships awaiting loading or unloading. Have a good organization of the terminal's container storage area requires a special effort. Containers search times can be considerable and lead to delays causing financial penalties for terminal management operators. An optimal location for a container in a terminal is very important for companies because this operation reduces transportation costs. In this article, we propose an approach for the placement problem of hazardous containers through the description of an optimization model to solve and optimize available storage space. The interest of this work is to develop a computer tool to enable the identification of the best location of a hazardous container using the Genetic Algorithm.

Purpose.The main design requirements of the modernized cars are those allowing reduction of operating costs and improvement of economic efficiency of their use. Due to the relevance of this subject the work presents the complex of conducted research, which will allow in prospect to use the universal flatcar, converted according to the Technical specifications TU 3182-065-71390252-2911 for container service. The research includes: evaluation of strength, fatigue safety of the design elements and conformity assessment of the strength characteristics of the modernized flatcar elements to the regulatory documents.Methodology.The analysis covers the use of specialized and universal rolling stock for transportation of containers, as well as the issues of modernization of universal cars. The strength of the flatcar bearing structure is evaluated based on the complex of numerical and experimental studies. The experimental part includes the static, repair load and impact tests. The car strength qualities and the fatigue safety are evaluated on the basis of calculated and experimental data.Findings.The conducted static, repair load and impact tests, given the quasi-static longitudinal forces, allowed the evaluation of the car structure strength according to the regulatory documents. The calculated and experimental data make it possible to assess the fatigue safety of structural elements. The present work is completed by obtaining the results that allow performing reasonable re-equipment of universal flatcars with baseplates with fitting stops to fasten the containers.Originality.The results of calculated and experimental studies showed that the modernized flatcars meet the conditions of strength and have sufficient fatigue safety factor. This makes it possible to recommend the re-equipment of universal flatcars for all car-repair enterprises that submitted the prototypes.Practical value.The complex of works was conducted that demonstrated the possibility of re-equipment of universal flatcars for the transportation of large containers. Part of the car fleet, kept in reserve, with little material costs can be transferred to the operational fleet. The modernization – re-equipment of universal flatcars with stationary specialized devices will increase the car usability coefficient.

Nowadays, in order to keep track of the fast changing requirements of Internet applications, auto-scaling is used as an essential mechanism for adapting the number of provisioned resources to the resource demand. The straightforward approach is to deploy a set of common and opensource single-service auto-scalers for each service independently. However, this deployment leads to problems such as bottleneck-shifting and increased oscillations. Existing auto-scalers that scale applications consisting of multiple services are kept closed-source. To face these challenges, we first survey existing auto-scalers and highlight current challenges. Then, we introduce Chamulteon, a redesign of our previously introduced mechanism, which can scale applications consisting of multiple services in a coordinated manner. We evaluate Chamulteon against four different well-cited auto-scalers in four sets of measurement-based experiments where we use diverse environments (VM vs. Docker), real-world traces, and vary the scale of the demanded resources. Overall, Chamulteon achieves the best auto-scaling performance based on established user-oriented and endorsed elasticity metrics.

The study deals with a shunting impact of 13-4085 wagon platform to a TK25 tank container. In order to define the loads acting on the container frame, 3D computer models of the 1СС size container is designed, both version – for liquid and dry material transport. The virtual model consider a longitudinal force from the wagon automatic coupling, vertical and horizontal forces from the wagon – containers interaction and also vertical forces of the container fittings. For checking adequacy of the created models, the container acceleration values was obtained in two scenarios. In the first case, when fittings are aligned relative to the fitting spots of a wagon, thus without possibility of relative movement. In the second case, movement within clearance of fitting spots is possible. Mathematical models of dynamic loads on these structures were solved by MathCad. Differential equations were solved by the Runge – Kutta method. Adequacy of the created models was checked with an F-test. On the basis of conducted calculation can be conclude that the hypothesis on the adequacy of the models was not rejected.

Traditional character segmentation has low accuracy for container scene text recognition. Convolutional recurrent neural network (CRNN) and connectionist text proposal network (CTPN) methods cannot extract container text features effectively. This paper proposes a novel Container Text Detection and Recognition Network (CTDRNet) for accurately detecting and recognizing container scene text. The CTDRNet consists of three components: (1) CTDRNet text detection enables to improve detection accuracy for single words; (2) CTDRNet text recognition has faster convergence speed and detection accuracy; (3) CTDRNet post-processing improves detection and recognition accuracy. In the end, the CTDRNet is implemented and evaluated with an accuracy of 96% and processing rate of 2.5 fps.

Operating system (OS) containers are becoming increasingly popular in cloud computing for improving productivity and code portability. However, existing deployment scheduling solutions mainly treat each container deployment as an independent request, and focus on the single aspect of resource utilization or load balancing, or work on homogeneous clusters. In this paper, we propose a new container deployment algorithm to satisfy multiple objectives on heterogeneous clusters. We analyze the deployment requirements of container-based infrastructure and formulate the deployment problem as a vector bin packing problem with heterogeneous bins. We focus on three objectives: multi-resource guarantee, load balancing, and dependency awareness. The goal of the proposed algorithm is to improve the tradeoff between load balancing and dependency awareness with multi-resource guarantees. Based on the algorithm, we implement a prototype scheduler to deploy containers on heterogeneous clusters. We evaluate our scheduler over a wide range of workload scenarios by simulation, which shows that our scheduler significantly outperforms existing schedulers of the container orchestration platforms.

Autoscaling of containers can leverage performance measures from the different layers of the computational stack. This paper investigate the problem of selecting the most appropriate performance measure to activate auto-scaling actions aiming at guaranteeing QoS constraints. First, the correlation between absolute and relative usage measures and how a resource allocation decision can be influenced by them is analyzed in different workload scenarios. Absolute and relative measures could assume quite different values. The former account for the actual utilization of resources in the host system, while the latter account for the share that each container has of the resources used. Then, the performance of a variant of Kubernetes’ auto-scaling algorithm, that transparently uses the absolute usage measures to scale-in/out containers, is evaluated through a wide set of experiments. Finally, a detailed analysis of the state-of-the-art is presented. open access

The internet of things (IoT) is essential for the implementation of applications and services that require the ability to sense the surrounding environment through sensors and modify it through actuators. However, IoT devices usually have limited computing capabilities and hence are not always sufficient to directly host resource-intensive services. Fog computing, which extends and complements the cloud, can support the IoT with computing resources and services that are deployed close to where data are sensed and actions need to be performed. Virtualisation is an essential feature in the cloud as in the fog, and containers have been recently getting much popularity to encapsulate fog services. Besides, container migration among fog nodes may enable several emerging use cases in different IoT domains (e.g., smart transportation, smart industry). In this paper, we first report container migration use cases in the fog and discuss containerisation. We then provide a comprehensive overview of the state-of-the-art migration techniques for containers, i.e., cold, pre-copy, post-copy, and hybrid migrations. The main contribution of this work is the extensive performance evaluation of these techniques that we conducted over a real fog computing testbed. The obtained results shed light on container migration within fog computing environments by clarifying, in general, which migration technique might be the most appropriate under certain network and service conditions.

International audience; The transition from virtual machine-based infrastructures to container-based ones brings the promise of swift and efficient software deployment in large-scale computing infrastructures. However, in fog computing environments which are often made of very small computers such as Raspberry PIs, deploying even a very simple Docker container may take multiple minutes. We demonstrate that Docker makes inefficient usage of the available hardware resources, essentially using different hardware subsystems (network bandwidth, CPU, disk I/O) sequentially rather than simultaneously. We therefore propose three optimizations which, once combined, reduce container deployment times by a factor up to 4. These optimizations also speed up deployment time by about 30% in datacenter-grade servers.

In this paper we propose a fuzzy neural network prediction approach based onmetaheuristics for container flow forecasting. The approach uses fuzzy if–then rulesfor selection between two different heuristics for developing neural network architecture, simulated annealing and genetic algorithm, respectively. These non-parametricmodels are compared with traditional parametric ARIMA technique. Timeseries composed from monthly container traffic observations for Port of Barcelonaare used for model developing and testing. Models are compared based on the mostimportant criteria for performance evaluation and for each of the data sets (totalcontainertraffic, loaded, unloaded, transit and empty) the appropriate model is selected.