Performance Exploration of Virtualization Systems

2016

Virtualization is used in data center and cloud environments to decouple applications from the hardware they run on. Hardware virtualization and operating system level virtualization are two prominent technologies that enable this. Containers, which use OS virtualization, have recently surged in interest and deployment. In this paper, we study the differences between the two virtualization technologies. We compare containers and virtual machines in large data center environments along the dimensions of performance, manageability and software development. We evaluate the performance differences caused by the different virtualization technologies in data center environments where multiple applications are running on the same servers (multi-tenancy). Our results show that co-located applications can cause performance interference, and the degree of interference is higher in the case of containers for certain types of workloads. We also evaluate differences in the management frameworks ...

Existing benchmarking methods are time consuming processes as they typically benchmark the entire Virtual Machine (VM) in order to generate accurate performance data, making them less suitable for real-time analytics. The research in this paper is aimed to surmount the above challenge by presenting DocLite-Docker Container-based Lightweight benchmarking tool. DocLite explores lightweight cloud benchmarking methods for rapidly executing benchmarks in near real-time. DocLite is built on the Docker container technology, which allows a user-defined memory size and number of CPU cores of the VM to be benchmarked. The tool incorporates two benchmarking methods-the first referred to as the native method employs containers to benchmark a small portion of the VM and generate performance ranks, and the second uses historic benchmark data along with the native method as a hybrid to generate VM ranks. The proposed methods are evaluated on three use-cases and are observed to be up to 91 times faster than benchmarking the entire VM. In both methods, small containers provide the same quality of rankings as a large container. The native method generates ranks with over 90% and 86% accuracy for sequential and parallel execution of an application compared against benchmarking the whole VM. The hybrid method did not improve the quality of the rankings significantly.

Cloud Computing is one of the main advancements that offer a promising future. One of the fundamental tools which are making the Cloud a reality is virtualization. Cloud computing provides various types of services such as Software as a service, PaaS, IaaS etc. PaaS vendors confront challenges in proficiently furnishing services with the huge growth of their offerings and hosting the distributed Applications. Hypervisor based virtualization comes with higher resource and operational overhead due to an extra level of abstraction. Virtual machine gives high service downtime when the application is updated to the new version. Containers have a favorable position over virtual machines because of performance enhancements and reduced start up time. Containers are particularly strong in managing PaaS clouds, such as application building, shipping and orchestration. Lightweight containers being the recent trends in the cloud scenario are also evolving as an important stage in PaaS service with the rise of Docker. In this paper, we have presented the performance overheads and application performance comparison among traditional hypervisor and container based virtualization. NOMENCLATURE PaaS-Platform as a service, IaaS: Infrastructure as a service, SaaS-Software as a service, VM-Virtual machine,

Proceedings of the 5th International Conference on Cloud Computing and Services Science, 2015

Experts argue that the resource demands of High Performance Computing (HPC) clusters request bare-metal installations. The performance loss of container virtualization is minimal and close to bare-metal, but in comparison has many advantages, like ease of provisioning. This paper presents the use of the newly adopted container technology and its multiple conceptional advantages for HPC, compared to traditional bare-metal installations or the use of VMs. The setup based on Docker (Docker, 2015) shows a possible use in private HPC sites or public clouds as well. The paper ends with a performance comparison of a FEA job run both bare-metal and using Docker and a detailed risk analysis of Docker installations in a multi-tenant environment, as HPC sites usually are.

Proceedings of the 10th International Conference on Cloud Computing and Services Science

The co-location of containers on the same host leads to significant performance concerns in the multi-tenant environment such as the cloud. These concerns are raised due to an attempt to maximize host resource utilization by increasing the number of containers running on the same host. The selection of a container runtime becomes critical in the case of strict performance requirements. In the scope of the study, two commonly used runtimes were evaluated: containerd (industry standard) and CRI-O (reference implementation of the CRI) on two different Open Container Initiative (OCI) runtimes: runc (default for most container runtimes) and gVisor (highly secure alternative to runc). Evaluation aspects of container runtimes include the performance of running containers, the performance of container runtime operations, and scalability. A tool called Touch-Stone was developed to address these evaluation aspects. The tool uses the CRI standard and is pluggable into any Kubernetes-compatible container runtime. Performance results demonstrate the better performance of containerd in terms of CPU usage, memory latency and scalability aspects, whereas file system operations (in particular, write operations) are performed more efficiently by CRI-O.

Service Science: Research and Innovations in the Service Economy, 2012

Nowadays, virtualization solutions are gaining increasing importance. By enabling the sharing of physical resources, thus making resource usage more efficient, they promise energy and cost savings. Additionally, virtualization is the key enabling technology for Cloud Computing and server consolidation. However, resource sharing and other factors have direct effects on system performance, which are not yet well-understood. Hence, performance prediction and performance management of services deployed in virtualized environments like public and private Clouds is a challenging task. Because of the large variety of virtualization solutions, a generic approach to predict the performance overhead of services running on virtualization platforms is highly desirable. In this paper, we present a methodology to quantify the influence of the identified performance-relevant factors based on an empirical approach using benchmarks. We show experimental results on two popular stateof-the-art virtualization platforms, Citrix XenServer 5.5 and VMware ESX 4.0, as representatives of the two major hypervisor architectures. Based on these results, we propose a basic, generic performance prediction model for the two different types of hypervisor architectures. The target is to predict the performance overhead for executing services on virtualized platforms.

2021

Virtual Machines (VMs) are used extensively in cloud computing. The underlying hypervisor allows hardware resources to be split into multiple virtual units which enhances resource utilization. However, VMs with traditional architecture introduce heavy overhead and reduce application performance. Containers have been introduced to overcome this drawback, yet such a solution raises security concerns due to poor isolation. Lightweight hypervisors have been leveraged to strike a balance between performance and isolation. However, there has been no comprehensive performance comparison among them. To identify the best fit use case, we investigate the performance characteristics of Docker container, Kata containers, gVisor, Firecracker and QEMU/KVM by measuring the performance on disk storage, main memory, CPU, network, system call and startup time. In addition, we evaluate their performance of running Nginx web server and MySQL database management system. We use QEMU/KVM as an example of ...

2017

In recent years, System Virtualization became a fundamental IT tool, whether it is type-2/hosted virtualization, mostly exploited by end-users in their personal computers, or type-1/bare metal, well established in IT departments and thoroughly used in modern datacenters as the very foundation of cloud computing. Though bare metal virtualization is meant to be deployed on server-grade hardware (for performance, stability and reliability reasons), properly configured desktop-class systems are often used as virtualization “servers”, due to their attractive performance/cost ratio. This paper presents the results of a study conducted on such systems, about the performance of Windows 10 and Ubuntu Server 16.04 guests, when deployed in what we believe are the type-1 platforms most in use today: VMware ESXi, Citrix XenServer, Microsoft Hyper-V, and KVM-based (represented by oVirt and Proxmox). Performance is measured using three synthetic benchmarks: PassMark for Windows, UnixBench for Ubun...

—Virtualization is a mature technology which has shown to provide computing resource and cost optimization while enabling consolidation, isolation and hardware abstraction through the concept of virtual machine. Recently, by sharing the operating system resources and simplifying the deployment of applications, containers are getting a more and more popular alternative to virtualization for specific use cases. As a result, today these two technologies are competing to provide virtual instances for cloud computing, Network Functions Virtualization (NFV), High Performance Computing (HPC), avionic and automotive platforms. In this paper, the performance of the most important open source hypervisor (KVM and Xen) and container (Docker) solutions are compared on the ARM architecture, which is rapidly emerging in the server world. The extensive system and Input/Output (I/O) performance measurements included in this paper show a slightly better performance for containers in CPU bound workloads and request/response networking; conversely, thanks to their caching mechanisms, hypervisors perform better in most disk I/O operations and TCP streaming benchmark.

Springer Arabian Journal for Science and Engineering, SCI Indexed, 2021

Containers emerged as a lightweight alternative to virtual machines that offer better microservice architecture support. They are widely used by organizations to deploy their increasingly diverse workloads derived from modern applications such as big data, IoT, and edge/fog computing in either proprietary clusters or private, public cloud data centers. With the growing interest in container-based virtualization technologies, the requirement to explore the deployment and orchestration of clusters of containers has become a central research problem. Although progress has been made to study containerization, systematic consolidation of the existing literature with a summative evaluation is still missing. To fill this gap, in this paper, we first taxonomically classify the existing research studies on the performance comparison between hypervisor and container technology and then analyze state-of-the-art for container cluster management orchestration systems, its performance monitoring tools, and finally future research trends. This results in a better understanding of container technology with attention to provide summative analysis in terms of (i) how much performance overhead is generated by a hypervisor compared to container-based virtualization, (ii) which container technology is suited for a cloud application deployment based on the type of benchmark executing, (iii) how to provide management of containers deployed in a cluster environment, (iv) container performance monitoring tools, and (v) finally emerging concerns for future research directions.