Lightweight Virtualization Technology Research Articles

REDB: Real-time enhancement of Docker containers via memory bank partitioning in multicore systems

Docker is a lightweight virtualization technology adopted by numerous companies to develop, deploy, and manage applications. However, All Docker containers running on the same system share system resources, leading to performance interference among them. This interference adversely affects the predictability of program response times within the containers. Thus, shared resource isolation between containers has become an important issue. Efforts have been made to improve isolation across various dimensions, such as file systems, I/O, etc. However, consideration of the effect of memory bank interference remains open. We propose a novel approach, REDB, as an effective solution to achieve memory isolation among containers, thereby enhancing the real-time performance of Docker containers. The fundamental strategy is partitioning DRAM banks and allocating specific banks to particular containers. The results reveal that the slowdown ratios of SPEC CPU2006 benchmarks show an average improvement of 7.7% (up to 18.4%) for Docker with REDB. We utilize the Cyclictest tool to characterize the Docker container latency, and our evaluations indicate that the maximum latency, a critical indicator of real-time performance, decreases significantly from 564 us for Docker without bank partitioning to just 26 us for Docker with REDB, resulting in an impressive reduction of 95.3%.

End-to-End Emulation of LoRaWAN Architecture and Infrastructure in Complex Smart City Scenarios Exploiting Containers.

In a LoRaWAN network, the backend is generally distributed as Software as a Service (SaaS) based on container technology, and recently, a containerized version of the LoRaWAN node stack is also available. Exploiting the disaggregation of LoRaWAN components, this paper focuses on the emulation of complex end-to-end architecture and infrastructures for smart city scenarios, leveraging on lightweight virtualization technology. The fundamental metrics to gain insights and evaluate the scaling complexity of the emulated scenario are defined. Then, the methodology is applied to use cases taken from a real LoRaWAN application in a smart city with hundreds of nodes. As a result, the proposed approach based on containers allows for the following: (i) deployments of functionalities on diverse distributed hosts; (ii) the use of the very same SW running on real nodes; (iii) the simple configuration and management of the emulation process; (iv) affordable costs. Both premise and cloud servers are considered as emulation platforms to evaluate the resource request and emulation cost of the proposed approach. For instance, emulating one hour of an entire LoRaWAN network with hundreds of nodes requires very affordable hardware that, if realized with a cloud-based computing platform, may cost less than USD 1.

Machine Learning Systems are Bloated and Vulnerable

Today's software is bloated with both code and features that are not used by most users. This bloat is prevalent across the entire software stack, from operating systems and applications to containers. Containers are lightweight virtualization technologies used to package code and dependencies, providing portable, reproducible and isolated environments. For their ease of use, data scientists often utilize machine learning containers to simplify their workflow. However, this convenience comes at a cost: containers are often bloated with unnecessary code and dependencies, resulting in very large sizes. In this paper, we analyze and quantify bloat in machine learning containers. We develop MMLB, a framework for analyzing bloat in software systems, focusing on machine learning containers. MMLB measures the amount of bloat at both the container and package levels, quantifying the sources of bloat. In addition, MMLB integrates with vulnerability analysis tools and performs package dependency analysis to evaluate the impact of bloat on container vulnerabilities. Through experimentation with 15 machine learning containers from TensorFlow, PyTorch, and Nvidia, we show that bloat accounts for up to 80% of machine learning container sizes, increasing container provisioning times by up to 370% and exacerbating vulnerabilities by up to 99%.

Toward Optimal Load Prediction and Customizable Autoscaling Scheme for Kubernetes

Most enterprise customers now choose to divide a large monolithic service into large numbers of loosely-coupled, specialized microservices, which can be developed and deployed separately. Docker, as a light-weight virtualization technology, has been widely adopted to support diverse microservices. At the moment, Kubernetes is a portable, extensible, and open-source orchestration platform for managing these containerized microservice applications. To adapt to frequently changing user requests, it offers an automated scaling method, Horizontal Pod Autoscaler (HPA), that can scale itself based on the system’s current workload. The native reactive auto-scaling method, however, is unable to foresee the system workload scenario in the future to complete proactive scaling, leading to QoS (quality of service) violations, long tail latency, and insufficient server resource usage. In this paper, we suggest a new proactive scaling scheme based on deep learning approaches to make up for HPA’s inadequacies as the default autoscaler in Kubernetes. After meticulous experimental evaluation and comparative analysis, we use the Gated Recurrent Unit (GRU) model with higher prediction accuracy and efficiency as the prediction model, supplemented by a stability window mechanism to improve the accuracy and stability of the prediction model. Finally, with the third-party custom autoscaling framework, Custom Pod Autoscaler (CPA), we packaged our custom autoscaling algorithm into a framework and deployed the framework into the real Kubernetes cluster. Comprehensive experiment results prove the feasibility of our autoscaling scheme, which significantly outperforms the existing Horizontal Pod Autoscaler (HPA) approach.

Multi agent deep reinforcement learning for resource allocation in container‐based clouds environments

AbstractVirtualization enables the deployment of several virtual servers on the same physical layer, critical component of the cloud. As cloud services advance, more apps that use repositories are developed, which adds to the overburden. Containers have evolved into the most reliable and lightweight virtualization technology for cloud services thanks to their flexible sorting, mobility, and scalability. In container‐based clouds, containers can potentially cut data centre energy usage more than virtual machines (VMs) do. Containers are less energy intensive than VMs. Resource allocation is the most prevalent method in cloud systems. However, resource allocation in container‐based clouds (RAC) is innovative and complicated due to its two‐level architecture. This includes the pairing of virtual machines and physical computers with containers. In cloud container services, planner components are essential. This lowers expenses while improving the performance and variety of workloads using cloud resources. The cloud infrastructure resource allocation framework is gaining popularity since it is energy‐efficient and focuses on cloud data management to maximize income and minimize costs. In this paper, we proposed a deep learning‐based architecture capable of achieving high data centre energy efficiency and preventing Service Level Agreement (SLA) violations from deploying green cloud resources. This research describes a hybrid optimum and multi‐agent deep reinforcement learning (MADRL) technique for dynamic task scheduling (DTS) in a container cloud environment. The MADRL‐DTS model for the RAC problem considers VM overheads, VM types, and an affinity restriction. Then, to address the RAC issue, we develop a DTS hyper‐heuristic technique. MADRL‐RAC may give allocation rules by recognizing workload trends and VM types from previous workload traces. Compared to modern procedures, the results demonstrate a significant reduction in energy consumption. The evaluation for energy‐efficient resource allocation is tested in several virtualized environments to get a high power usage effectiveness and CPU usage.

Containerized Architecture Performance Analysis for IoT Framework Based on Enhanced Fire Prevention Case Study: Rwanda.

Nowadays, building infrastructures are pushed to become smarter in response to desires for the environmental comforts of living. Enhanced safety upgrades have begun taking advantage of new, evolving technologies. Normally, buildings are configured to respond to the safety concerns of the occupants. However, advanced Internet of Things (IoT) techniques, in combination with edge computing with lightweight virtualization technology, is being used to improve users’ comfort in their homes. It improves resource management and service isolation without affecting the deployment of heterogeneous hardware. In this research, a containerized architectural framework for support of multiple concurrent deployed IoT applications for smart buildings was proposed. The prototype developed used sensor networks as well as containerized microservices, centrally featuring the DevOps paradigm. The research proposed an occupant counting algorithm used to check occupants in and out. The proposed framework was tested in different academic buildings for data acquisition over three months. Different deployment architectures were tested to ensure the best cases based on efficiency and resource utilization. The acquired data was used for prediction purposes to aid occupant prediction for safety measures as considered by policymakers.

Ant Colony Optimization-based Light Weight Container (ACO-LWC) Algorithm for Efficient Load Balancing

Container technology is the latest lightweight virtualization technology which is an alternate solution for virtual machines. Docker is the most popular container technology for creating and managing Linux containers. Containers appear to be the most suitable medium for use in dynamic development, packaging, shipping and many other information technology environments. The portability of the software through the movement of containers is appreciated by businesses and IT professionals. In the docker container, one or more processes may run simultaneously. The main objective of this work is to propose a new algorithm called Ant Colony Optimization-based Light Weight Container (ACO-LWC) load balancing scheduling algorithm for scheduling various process requests. This algorithm is designed such that it shows best performance in terms of load balancing. The proposed algorithm is validated by comparison with two existing load balancing scheduling algorithms namely, least connection algorithm and round robin algorithm. The proposed algorithm is validated using metrics like response time (ms), mean square error (MSE), node load, largest Transactions Per Second (TPS) of cluster (fetches/sec), average response time for each request (ms) and run time (s). Quantitative analysis show that the proposed ACO-LWC scheme achieves best performance in terms of all the metrics compared to the existing algorithms. In particular, the response time for least connection, round robin and the proposed ACO-LWC algorithm are 58, 60 and 48 ms respectively when 95% requests are finished. Similarly, the error for scheduling 120 requests using least connection, round robin and the proposed ACO-LWC algorithm are 0.15, 0.11 and 0.06 respectively.

Containers vs. virtual machines: performance comparison

Server virtualization is a technological innovation widely used in information technology (IT) companies. Virtualization provides a platform to run different operating system services in the cloud. It makes it easy to build multiple virtual machines on a single basic physical machine, either in the form of hypervisors or containers. To host many microservices applications, emerging technology has introduced a model consisting of different operations performed by smaller individual deployed services. Therefore, the demand for low-cost virtualization techniques is developing rapidly. There are many lightweight virtualization technologies; Docker is one of them, which is an open-source platform. This technology enables developers and system administrators to build, create, and run applications using the Docker engine. This document provides performance evaluation of Docker containers and virtual machines using standard benchmark tools such as Sysbench, Phoronix, and Apache, including CPU performance, memory performance, storage read/write performance, load test and measurement of operating speed.

Elastic deployment of container clusters across geographically distributed cloud data centers for web applications

AbstractContainers such as Docker provide a lightweight virtualization technology. They have gained popularity in developing, deploying and managing applications in and across Cloud platforms. Container management and orchestration platforms such as Kubernetes run application containers in virtual clusters that abstract the overheads in managing the underlying infrastructures to simplify the deployment of container solutions. These platforms are well suited for modern web applications that can give rise to geographic fluctuations in use based on the location of users. Such fluctuations often require dynamic global deployment solutions. A key issue is to decide how to adapt the number and placement of clusters to maintain performance, whilst incurring minimum operating and adaptation costs. Manual decisions are naive and can give rise to: over‐provisioning and hence cost issues; improper placement and performance issues, and/or unnecessary relocations resulting in adaptation issues. Elastic deployment solutions are essential to support automated and intelligent adaptation of container clusters in geographically distributed Clouds. In this article, we propose an approach that continuously makes elastic deployment plans aimed at optimizing cost and performance, even during adaptation processes, to meet service level objectives (SLOs) at lower costs. Meta‐heuristics are used for cluster placement and adjustment. We conduct experiments on the Australia‐wide National eResearch Collaboration Tools and Resources Research Cloud using Docker and Kubernetes. Results show that with only a 0.5 ms sacrifice in SLO for the 95th percentile of response times we are able to achieve up to 44.44% improvement (reduction) in cost compared to a naive over‐provisioning deployment approach.

Evolving Container to Unikernel for Edge Computing and Applications in Process Industry

Industry 4.0 promotes manufacturing and process industry towards digitalization and intellectualization. Edge computing can provide delay-sensitive services in industrial processes to realize intelligent production. Lightweight virtualization technology is one of the key elements of edge computing, which can implement resource management, orchestration, and isolation services without considering heterogenous hardware. It has revolutionized software development and deployment. The scope of this review paper is to present an in-depth analysis of two such technologies, Container and Unikernel, for edge computing. We discuss and compare their applicability in terms of migration, security, and orchestration for edge computing and industrial applications. We describe their performance indexes, evaluation methods and related findings. We then discuss their applications in industrial processes. To promote further research, we present some open issues and challenges to serve as a road map for both researchers and practitioners in the areas of Industry 4.0, industrial process automation, and advanced computing.

Extending Kubernetes Clusters to Low-Resource Edge Devices Using Virtual Kubelets

In recent years, containers have gained popularity as a lightweight virtualization technology. This rise in popularity has gone hand in hand with the adoption of microservice architectures, mostly thanks to the scalable, ethereal, and isolated nature of containers. More recently, edge devices have become powerful enough to be able to run containerized microservices, while remaining flexible enough in terms of size and power to be deployed almost anywhere. This has triggered research into several container placement strategies involving edge networks, leading to concepts such as osmotic computing. While these container placement strategies are optimal in terms of workload placement, current container orchestrators are often not suitable for running on edge devices due to their high resource requirements. In this article, FLEDGE is presented as a Kubernetes-compatible container orchestrator based on Virtual Kubelets, aimed primarily at container orchestration on low-resource edge devices. Several aspects of low-resource container orchestration are examined, such as the choice of container runtime and how to realize container networking. A number of evaluations are performed to determine how FLEDGE compares to Kubernetes and K3S in terms of resource requirements, showing that it needs around 60MiB memory and 78MiB storage to run on a Raspberry Pi 3, including all dependencies, which is significantly less than both studied alternatives.

Improving the efficiency of HPC data movement on container-based virtual cluster

Today, lightweight virtualization technologies have been widely deployed on data centers and HPC clusters to provide highly efficient and elastic resource provisioning. Virtualization has also been extended to the I/O stack in operating system. For example, virtual switch has become the primary provider of I/O services for data movement among various light-weight virtual machines, such as Docker and Kubernetes. However, I/O stack virtualization introduces performance degradation and scalability bottleneck to the data movements of HPC computing framework, such as MPI based collective data movements and bursty asynchronous data movements. In order to study the bottleneck, we quantify and analyze the performance degradation involving with HPC data movements on virtual clusters. Then, we design a set of two-stage methods to proactively adapt the virtual network and data movement procedures. This can enhance the performance of HPC collective data movements by up to 3$$\times $$. Meanwhile, a cross-layer middleware is designed to improve the performance and scalability of bursty asynchronous data movements. Our evaluation shows that it can improve the performance of real scientific application by 34.6%.

The state‐of‐the‐art in container technologies: Application, orchestration and security

SummaryContainerization is a lightweight virtualization technology enabling the deployment and execution of distributed applications on cloud, edge/fog, and Internet‐of‐Things platforms. Container technologies are evolving at the speed of light, and there are many open research challenges. In this paper, an extensive literature review is presented that identifies the challenges related to the adoption of container technologies in High Performance Computing, Big Data analytics, and geo‐distributed (Edge, Fog, Internet‐of‐Things) applications. From our study, it emerges that performance, orchestration, and cyber‐security are the main issues. For each challenge, the state‐of‐the‐art solutions are then analyzed. Performance is related to the assessment of the performance footprint of containers and comparison with the footprint of virtual machines and bare metal deployments, the monitoring, the performance prediction, the I/O throughput improvement. Orchestration is related to the selection, the deployment, and the dynamic control of the configuration of multi‐container packaged applications on distributed platforms. The focus of this work is on run‐time adaptation. Cyber‐security is about container isolation, confidentiality of containerized data, and network security. From the analysis of 97 papers, it came out that the state‐of‐the‐art is more mature in the area of performance evaluation and run‐time adaptation rather than in security solutions. However, the main unsolved challenges are I/O throughput optimization, performance prediction, multilayer monitoring, isolation, and data confidentiality (at rest and in transit).

Standalone containers with ATLAS offline software

This paper describes the deployment of the offline software of the ATLAS experiment at LHC in containers for use in production workflows such as simulation and reconstruction. To achieve this goal we are using Docker and Singularity, which are both lightweight virtualization technologies that can encapsulate software packages inside complete file systems. The deployment of offline releases via containers removes the interdependence between the runtime environment needed for job execution and the configuration of the computing nodes at the sites. Docker or Singularity would provide a uniform runtime environment for the grid, HPCs and for a variety of opportunistic resources. Additionally, releases may be supplemented with a detector’s conditions data, thus removing the need for network connectivity at computing nodes, which is normally quite restricted for HPCs. In preparation to achieve this goal, we have built Docker and Singularity images containing single full releases of ATLAS software for running detector simulation and reconstruction jobs in runtime environments without a network connection. Unlike similar efforts to produce containers by packing all possible dependencies of every possible workflow into heavy images (≈ 200GB), our approach is to include only what is needed for specific workflows and to manage dependencies efficiently via software package managers. This approach leads to more stable packaged releases where the dependencies are clear and the resulting images have more portable sizes ( 16GB). In an effort to cover a wider variety of workflows, we are deploying images that can be used in raw data reconstruction. This is particularly challenging due to the high database resource consumption during the access to the experiment’s conditions payload when processing data. We describe here a prototype pipeline in which images are provisioned only with the conditions payload necessary to satisfy the jobs’ requirements. This database-on-demand approach would keep images slim, portable and capable of supporting various workflows in a standalone fashion in environments with no network connectivity.

Performance Evaluation of Docker Container and Virtual Machine

Abstract Server virtualization is a technological innovation broadly used in IT enterprises. Virtualization provides a platform to run different services of operating systems on the cloud. It facilitates to build multiple virtual machines on a single basic physical machine either in the form of hypervisors or containers. To host many microservice applications, the emergent technology has introduced a model which consists of different operations performed by smaller individual deployed services. Thus, the demand for low-overhead virtualization technique is rapidly developing. There are many lightweight virtualization technologies; docker is one among them, which is an open-source platform. This technology allows developers and system admins to build, create, and run applications using docker engine. This paper provides the performance evaluation of Docker containers and virtual machines using standard benchmark tools such as Sysbench, Phoronix, and Apache benchmark, which include CPU performance, Memory throughput, Storage read/write performance, load test, and operation speed measurement.

A placement architecture for a container as a service (CaaS) in a cloud environment

Unlike a traditional virtual machine (VM), a container is an emerging lightweight virtualization technology that operates at the operating system level to encapsulate a task and its library dependencies for execution. The Container as a Service (CaaS) strategy is gaining in popularity and is likely to become a prominent type of cloud service model. Placing container instances on virtual machine instances is a classical scheduling problem. Previous research has focused separately on either virtual machine placement on physical machines (PMs) or container, or only tasks without containerization, placement on virtual machines. However, this approach leads to underutilized or overutilized PMs as well as underutilized or overutilized VMs. Thus, there is a growing research interest in developing a container placement algorithm that considers the utilization of both instantiated VMs and used PMs simultaneously.The goal of this study is to improve resource utilization, in terms of number of CPU cores and memory size for both VMs and PMs, and to minimize the number of instantiated VMs and active PMs in a cloud environment. The proposed placement architecture employs scheduling heuristics, namely, Best Fit (BF) and Max Fit (MF), based on a fitness function that simultaneously evaluates the remaining resource waste of both PMs and VMs. In addition, another meta-heuristic placement algorithm is proposed that uses Ant Colony Optimization based on Best Fit (ACO-BF) with the proposed fitness function. Experimental results show that the proposed ACO-BF placement algorithm outperforms the BF and MF heuristics and maintains significant improvement of the resource utilization of both VMs and PMs.

Assessing Lightweight Virtualization for Security-as-a-Service at the Network Edge

Accounting for the exponential increase in security threats, the development of new defense strategies for pervasive environments is acquiring an ever-growing importance. The expected avalanche of heterogeneous IoT devices which will populate our industrial factories and smart houses will increase the complexity of managing security requirements in a comprehensive way. To this aim, cloud-based security services are gaining notable impetus to provide security mechanisms according to Security-as-a-Service (SECaaS) model. However, the deployment of security applications in remote cloud data-centers can introduce several drawbacks in terms of traffic overhead and latency increase. To cope with this, Edge Computing can provide remarkable advantages avoiding long routing detours. On the other hand, the limited capabilities of edge node introduce potential constraints in the overall management. This paper focuses on the provisioning of virtualized security services in resource-constrained edge nodes by leveraging lightweight virtualization technologies. Our analysis aims at shedding light on the feasibility of container-based security solutions, thus providing useful guidelines towards the orchestration of security at the edge. Our experiments show that the overhead introduced by the containerization is very light.

Container Based Resource Management for Data Processing on IoT Gateways

IoT is featured by its diversity and heterogeneity in terms of hardware and software components shaping its systems. This in turn burdens the process of deploying and updating hardware components and their corresponding software elements. Moreover, these devices are constantly generating data and sending them to central cloud platforms for advanced processing, with billions of devices are anticipated to be connected in the coming years, this will lead to impact the network performance. To alleviate this concern, a recent tendency is to process these data locally close to their sources at the network edge. However, edge devices are likely to be resource-constrained, which in some cases limits their abilities to perform processing tasks. This work falls within the efforts attempt to provide a solution to enable an effective management of resources at the edge devices, as well as, facilitating the deployment and update of heterogeneous software modules. The proposed solution relies on containers, a lightweight virtualization technology. It takes advantage of some containers concepts, especially the orchestration concept, to enable task forwarding and resource sharing between IoT gateway devices within the same cluster. Our experimentation has shown promising results and revealed the potential of existing virtualization platforms.

Ant Colony Algorithm for Multi-Objective Optimization of Container-Based Microservice Scheduling in Cloud

In cloud architectures, the microservice model divides an application into a set of loosely coupled and collaborative fine-grained services. As a lightweight virtualization technology, the container supports the encapsulation and deployment of microservice applications. Despite a large number of solutions and implementations, there remain open issues that have not been completely addressed in the deployment and management of the microservice containers. An effective method for container resource scheduling not only satisfies the service requirements of users but also reduces the running overhead and ensures the performance of the cluster. In this paper, a multi-objective optimization model for the container-based microservice scheduling is established, and an ant colony algorithm is proposed to solve the scheduling problem. Our algorithm considers not only the utilization of computing and storage resources of the physical nodes but also the number of microservice requests and the failure rate of the physical nodes. Our algorithm uses the quality evaluation function of the feasible solutions to ensure the validity of pheromone updating and combines multi-objective heuristic information to improve the selection probability of the optimal path. By comparing with other related algorithms, the experimental results show that the proposed optimization algorithm achieves better results in the optimization of cluster service reliability, cluster load balancing, and network transmission overhead.

Combining containers and virtual machines to enhance isolation and extend functionality on cloud computing

Virtualization technology is the underlying element of cloud computing. Traditionally, cloud computing has employed virtual machines to distribute available resources and provide isolated environments among users. Multiple virtual machines, with their own operating system and services, can be deployed and run simultaneously on the same physical machine on the cloud infrastructure. Recently, a more lightweight virtualization technology is being rapidly adopted and it is based on containers. A key difference between virtual machines and containers, is that containers share the same underlying operating system.In this work, we studied the combination of these two virtualization technologies by running containers on top of virtual machines. This synergy aims to enhance containers’ main drawback, which is isolation and, among others, to simplify the system management and upgrade, and to introduce the new functionalities of containerized applications to virtual machines. The benefits of this method have been recognized by big cloud companies, which have been employing this approach for years.With this paper, we aimed to present the advantages of running containers on virtual machines and to explore how various virtualization techniques and configurations affect the performance of this method. Although, in bibliography there are a few papers that used this method partially to conduct experiments, there is not any research which considers this method from our integral aspect of view. In this study, we ran Docker containers and evaluated their performances, not only on KVM and XEN virtual machines, but also we ran Linux containers on Windows Server. We adopted an empirical approach, to quantify the performance overhead introduced by the additional virtualization layer of virtual machines, by executing various benchmarks and deploying real world applications as use cases. Furthermore, we presented for first time, how isolation is applied on virtual machines and containers, we evaluated different operating systems optimized to host containers, mechanisms of storing persistent data and, last but not least, we experimentally quantified the power and energy consumption overhead of containers running on virtual machines.