Virtualization Overhead Research Articles

Performance Evaluation of Parallel Haemodynamic Computations on Heterogeneous Clouds

The article presents performance evaluation of parallel haemodynamic flow computations on heterogeneous resources of the OpenStack cloud infrastructure. The main focus is on the parallel performance analysis, energy consumption and virtualization overhead of the developed software service based on ANSYS Fluent platform which runs on Docker containers of the private university cloud. The haemodynamic aortic valve flow described by incompressible Navier-Stokes equations is considered as a target application of the hosted cloud infrastructure. The parallel performance of the developed software service is assessed measuring the parallel speedup of computations carried out on virtualized heterogeneous resources. The performance measured on Docker containers is compared with that obtained by using the native hardware. The alternative solution algorithms are explored in terms of the parallel performance and power consumption. The investigation of a trade-off between the computing speed and the consumed energy is performed by using Pareto front analysis and a linear scalarization method.

VMobiDesk: Desktop Virtualization for Mobile Operating Systems

Smart mobile devices have significantly increased the popularity of Bring-Your-Own-Device (BYOD) at work, as they benefit people’s daily lives. However, BYOD comes with several challenging issues such as limited hardware capacity, frequent upgrades of applications, and security and privacy concerns. Virtual Mobile Infrastructure (VMI), a general framework that provides more reliable and secure solution for BYOD, has therefore been proposed. The key of VMI is to host a mobile Operating System (OS) on a remote cloud data center, and run mobile applications on it. However, VMI faces performance challenge as it needs to display the entire virtualized desktop on a mobile device while the real content of the desktop is on a remote server. To address the performance challenge, we design and implement a VMI named vMobiDesk on top of Android with optimized network transfer mechanisms and display virtualization. In particular, vMobiDesk focuses on virtualizing the display of Android desktops, redirecting users’ input events, providing audio support and remote camera. The experimental results show that vMobiDesk has low virtualization overhead, as well as enables mobile users to obtain good experiences with BYOD applications.

Virtio Front-End Network Driver for RTEMS Operating System

Virtualization technology can provide a transparent and efficient software development environment at the development phase of embedded systems. In addition, virtualization can be exploited to provide strong resource isolation between real-time applications. However, since virtualization incurs noticeable overheads, we need to minimize these to utilize virtualization technology for either a software development environment or a runtime environment of real-time embedded systems. In this letter, we particularly focus on virtualization overheads of network I/O and present an implementation of standardized (i.e., virtio) front-end network driver for a real-time operating system (RTOS). Although there have been several studies for virtio on general-purpose OS, our implementation is the first open-source virtio for an RTOS. The measurement results show that our implementation can improve the bandwidth and latency up to 168% and 52%, respectively. In addition, the memory usage can be saved up to 36%.

Disentangling the effects of photosynthetically active radiation and red to far-red ratio on plant photosynthesis under canopy shading: a simulation study using a functional-structural plant model.

Background and AimsShading by an overhead canopy (i.e. canopy shading) entails simultaneous changes in both photosynthetically active radiation (PAR) and red to far-red ratio (R:FR). As plant responses to PAR (e.g. changes in leaf photosynthesis) are different from responses to R:FR (e.g. changes in plant architecture), and these responses occur at both organ and plant levels, understanding plant photosynthesis responses to canopy shading needs separate analysis of responses to reductions in PAR and R:FR at different levels.MethodsIn a glasshouse experiment we subjected plants of woody perennial rose (Rosa hybrida) to different light treatments, and so separately quantified the effects of reductions in PAR and R:FR on leaf photosynthetic traits and plant architectural traits. Using a functional–structural plant model, we separately quantified the effects of responses in these traits on plant photosynthesis, and evaluated the relative importance of changes of individual traits for plant photosynthesis under mild and heavy shading caused by virtual overhead canopies.Key ResultsModel simulations showed that the individual trait responses to canopy shading could have positive and negative effects on plant photosynthesis. Under mild canopy shading, trait responses to reduced R:FR on photosynthesis were generally negative and with a larger magnitude than effects of responses to reduced PAR. Conversely, under heavy canopy shading, the positive effects of trait responses to reduced PAR became dominant. The combined effects of low-R:FR responses and low-PAR responses on plant photosynthesis were not equal to the sum of the separate effects, indicating interactions between individual trait responses.ConclusionsOur simulation results indicate that under canopy shading, the relative importance of plant responses to PAR and R:FR for plant photosynthesis changes with shade levels. This suggests that the adaptive significance of plant plasticity responses to one shading factor depends on plant responses to the other.

Dynamic Load Aware Scheduler of Map Reduce Tasks for Cloud Environments

Most of the current day applications are data and compute intensive which led to invention of technologies like Hadoop. Hadoop uses Map Reduce framework for parallel processing of big data applications using the computing resources of multiple nodes. Hadoop is designed for cluster environments and has few limitations when executed in cloud environments. Hadoop on cloud has become a common choice due to its easy establishment of infrastructure and pay as you use model. Hadoop performance on cloud infrastructures is affected by the virtualization overhead of cloud environment. The execution times of Hadoop on cloud can be improved if the virtual resources are effectively used to schedule the tasks by studying the resource usage characteristics of the tasks and resource availability of the nodes. The proposed work is to build a dynamic scheduler for Hadoop framework which can make scheduling decision dynamically based on job resource usage and node load. The results of the proposed work indicate an improvement of up to 23% in execution time of the Hadoop Map Reduce applications.

Backfilling the Grid with Containerized BOINC in the ATLAS computing

Virtualization is a commonly used solution for utilizing the opportunistic computing resources in the HEP field, as it provides a unified software and OS layer that the HEP computing tasks require over the heterogeneous opportunistic computing resources. However there is always performance penalty with virtualization, especially for short jobs which are always the case for volunteer computing tasks, the overhead of virtualization reduces the CPU efficiency of the jobs, hence it leads to low CPU efficiency of the jobs. With the wide usage of containers in HEP computing, we explore the possibility of adopting the container technology into the ATLAS BOINC project, hence we implemented a Native version in BOINC, which uses the Singularity container or direct usage of the Operating System of the host machines to replace VirtualBox. In this paper, we will discuss 1) the implementation and workflow of the Native version in the ATLAS BOINC; 2) the performance measurement of the Native version comparing to the previous virtualization version. 3) the limits and shortcomings of the Native version; 4) The practice and outcome of the Native version which includes using it in backfilling the ATLAS Grid Tier2 sites and other clusters, and to utilize the idle computers from the CERN computing centre.

GPUCloudSim: an extension of CloudSim for modeling and simulation of GPUs in cloud data centers

Recent years have witnessed an increasing growth in the usage of GPUs in cloud data centers. It is known that conventional virtualization techniques are not directly applicable to GPUs, making it a challenge to effectively take advantage of virtualization benefits. API remoting, full, para and hardware-assisted virtualization methods are adopted to empower VMs with GPU capabilities. With such a diversity in approaches, there is a need for a simulation environment to study the effectiveness of GPU virtualization techniques and evaluate GPU provisioning and scheduling policies in cloud data centers. In order to model and simulate GPU-enabled VMs in cloud data centers, this work proposes and describes a simulator architecture implemented as an extension of CloudSim. The extension eases up conducting experimental studies that otherwise need to be carried out in real cloud infrastructures. It includes models to simulate interference among co-running applications, the overhead of virtualization and power consumption of GPUs. To demonstrate the usefulness of our extension, we study NVIDIA GRID, a hardware-assisted GPU virtualization solution. We show that for situations where the number of VMs outperforms the number of hosts, the first-fit VM placement of VMware Horizon may not be effective. Instead, we suggest a first-fit increasing VM placement algorithm which increases the acceptance rate by 59%, shortens makespan by 25% and saves energy by 21%.

Big Data-Oriented PaaS Architecture with Disk-as-a-Resource Capability and Container-Based Virtualization

With the increasing adoption of Big Data technologies as basic tools for the ongoing Digital Transformation, there is a high demand for data-intensive applications. In order to efficiently execute such applications, it is vital that cloud providers change the way hardware infrastructure resources are managed to improve their performance. However, the increasing use of virtualization technologies to achieve an efficient usage of infrastructure resources continuously widens the gap between applications and the underlying hardware, thus decreasing resource efficiency for the end user. Moreover, this scenario is especially troublesome for Big Data applications, as storage resources are one of the most heavily virtualized, thus imposing a significant overhead for large-scale data processing. This paper proposes a novel PaaS architecture specifically oriented for Big Data where the scheduler offers disks as resources alongside the more common CPU and memory resources, looking forward to provide a better storage solution for the user. Furthermore, virtualization overheads are reduced to the bare minimum by replacing heavy hypervisor-based technologies with operating-system-level virtualization based on light software containers. This architecture has been deployed on a Big Data infrastructure at the CESGA supercomputing center, used as a testbed to compare its performance with OpenStack, a popular private cloud platform. Results have shown significant performance improvements, reducing the execution time of representative Big Data workloads by up to 4.5×.

Measuring Overhead of Concurrency and Virtual Memory

We present the methodology, as well as results of measurements and evaluation of overhead created by concurrency and virtual memory. A special measurement technique and testbed were used to obtain the most accurate data from the experiments. This technique is focused on the measurements of the overall performance degradation that is introduced by concurrency in the form of lightweight user-level threads on IA-32 processors. We have obtained and compared results of the experiments in an environment with and without enabled virtual memory to understand what loss of performance is caused by virtual memory in itself, and how it affects the overhead associated with concurrency. The results showed that overhead of concurrency outweighs virtual memory overhead and that there is a complex dependency between them. The article is published in the author’s wording.

Sugar

Modern personal computers have embraced increasingly powerful Graphics Processing Units (GPUs). Recently, GPU-based graphics acceleration in web apps (i.e., applications running inside a web browser) has become popular. WebGL is the main effort to provide OpenGL-like graphics for web apps and it is currently used in 53% of the top-100 websites. Unfortunately, WebGL has posed serious security concerns as several attack vectors have been demonstrated through WebGL. Web browsers» solutions to these attacks have been reactive: discovered vulnerabilities have been patched and new runtime security checks have been added. Unfortunately, this approach leaves the system vulnerable to zero-day vulnerability exploits, especially given the large size of the Trusted Computing Base of the graphics plane. We present Sugar, a novel operating system solution that enhances the security of GPU acceleration for web apps by design. The key idea behind Sugar is using a dedicated virtual graphics plane for a web app by leveraging modern GPU virtualization solutions. A virtual graphics plane consists of a dedicated virtual GPU (or vGPU) as well as all the software graphics stack (including the device driver). Sugar enhances the system security since a virtual graphics plane is fully isolated from the rest of the system. Despite GPU virtualization overhead, we show that Sugar achieves high performance. Moreover, unlike current systems, Sugar is able to use two underlying physical GPUs, when available, to co-render the User Interface (UI): one GPU is used to provide virtual graphics planes for web apps and the other to provide the primary graphics plane for the rest of the system. Such a design not only provides strong security guarantees, it also provides enhanced performance isolation.

Non-clairvoyant online scheduling of synchronized jobs on virtual clusters

Although virtualization technology is recently applied to next-generation distributed high-performance computing systems, theoretical aspects of scheduling jobs on these virtualized environments are not sufficiently studied, especially in online and non-clairvoyant cases. Virtualization of computing resources results in interference and virtualization overheads that negatively impact the load balancing objectives on commonly used cluster of multi-core physical machines. We present a technique for non-clairvoyant online scheduling of globally synchronized jobs, each of which spawns tasks to execute compute-intensive works. Our technique considers both load balancing of physical cores and per job synchronization cost minimization. We show that in the presence of arbitrary virtualization overheads, interference effects and synchronization cost, the problem can be reduced to an online unrelated parallel machine scheduling, which is solved using routing of virtual circuits. We present a new opportunity cost model to reduce the problem to the routing of virtual circuits and prove the effectiveness of our scheduling technique using mathematical analysis and simulative experiments.

Evaluation Methodology for Virtual Base Station Platforms in Radio Access Networks

Virtual base stations (BS) with scalable resources and flexible functions are necessary for the fifth-generation and future mobile systems. The virtualization platform is the fundamental component of virtual BSs and plays a very important role in determining BS performance. As BS virtualization platforms can be provided by both conventional BS manufactures and third-party software vendors, it is crucial to define unified methodology to evaluate the performance of different virtualization platforms. In this paper, for the first time in literature, we design a set of comprehensive performance metrics and propose evaluation methods for BS virtualization platforms. More specifically, there are two types of metrics, i.e., micro level and macro level metrics. Micro level metrics quantify the virtualization platforms’ basic performances including real-time performance and virtualization overhead. Macro level metrics quantify the platforms ability to support BS resource and function virtualization while satisfying the radio access network requirements. Correspondingly, two-level evaluation methodology is designed to collect micro and macro level metrics from different parts of a virtual BS by running proper workloads. Finally, using the proposed methodology, two representative virtualization platforms, i.e., KVM hypervisor and Super BS virtualization platform are evaluated. These experiments verify that the proposed methodology can effectively evaluate the performance of different platforms to provide a valuable reference for the selection of virtual BS platforms. Moreover, the proposed evaluation methodology is a significant step toward constructing an open ecosystem for radio access networks.

Performance Implications of Extended Page Tables on Virtualized x86 Processors

Managing virtual memory is an expensive operation, and becomes even more expensive on virtualized servers. Processing TLB misses on a virtualized x86 server requires a twodimensional page walk that can have 6x more page table lookups, hence 6x more memory references, than a native page table walk. Thus much of the recent research on the subject starts from the assumption that TLB miss processing in virtual environments is significantly more expensive than on native servers. However, we will show that with the latest software stack on modern x86 processors, most of these page table lookups are satisfied by internal paging structure caches and the L1/L2 data caches, and the actual virtualization overhead of TLB miss processing is a modest fraction of the overall time spent processing TLB misses. We show that even for the heaviest workloads, a welltuned application that uses large pages on a recent OS release with a modern hypervisor running on the latest x86 processors sees only minimal degradation from the additional overhead of the two-dimensional page walks in a virtualized server.

Towards a TrustZone-Assisted Hypervisor for Real-Time Embedded Systems

Virtualization technology starts becoming more and more widespread in the embedded space. The penalties incurred by standard software-based virtualization is pushing research towards hardware-assisted solutions. Among the existing commercial off-the-shelf technologies for secure virtualization, ARM TrustZone is attracting particular attention. However, it is often seen with some scepticism due to the dual-OS limitation of existing state-of-the-art solutions. This letter presents the implementation of a TrustZone-based hypervisor for real-time embedded systems, which allows multiple RTOS partitions on the same hardware platform. The results demonstrate that virtualization overhead is less than 2 percent for a 10 milliseconds guest-switching rate, and the system remains deterministic. This work goes beyond related work by implementing a TrustZone-assisted solution that allows the execution of an arbitrary number of guest OSes while providing the foundation to drive next generation of secure virtualization solutions for resource-constrained embedded devices.

I/O access frequency-aware cache method on KVM/QEMU

Together with the rapid development of IT technology, cloud computing has been considered as the next generation's computing infrastructure. One of the essential part of cloud computing is the virtual machine technology that enables to reduce the data center cost with better resource utilization. Especially, virtual desktop infrastructure (VDI) is receiving explosive attentions from IT markets because of its advantages of easier software management, greater data protection, and lower cost. However, sharing physical resources in VDI to consolidate multiple guest virtual machines (VMs) on a host has a tradeoff that can lead to significant I/O degradation. Optimizing I/O virtualization overhead is a challenging task because it needs to scrutinize multiple software layers between guest VMs and host where those VMs are executing. In this paper, we present a hypervisor-level cache, called hyperCache, which is possible to provide a shortcut in KVM/QEMU. It intercepts I/O requests in the hypervisor and analyses their I/O access patterns to select data retaining high access frequency. Also, it has a capability of maintaining the appropriate cache memory size by utilizing the cache block map. Our experimental results demonstrate that our method improves I/O bandwidth by up to 4.7x over the existing QEMU.

Autonomous Discovery and Management in Virtual Container Clusters

Global software stacks on scientific cluster computing resources are required to provide a homogeneous software environment which is typically inflexible. Efforts to integrate Virtual Machines (VMs), in order to abstract the software environment of various scientific applications, suffer from performance limitations and require systems administration expertise to maintain. However, the motivation is clear; in addition to increasing resource utilization, the burden of supporting new software installations on existing systems can be reduced. In this paper, we introduce the Virtual Container Cluster (VCC) that encapsulates a typical HPC software environment within Docker containers. The novel component cluster–watcher enables context aware discovery and configuration of the virtual cluster. Containers offer a lightweight alternative to VMs that more closely match the native performance, and presents a solution that is more accessible to customization by the average user. Combined with a Software Defined Networking (SDN) technology, the VCC enables dynamic features such as transparent scaling and spanning across multiple physical resources. Although SDN introduces an additional performance limitation, within the context of a parallel communication network, the benchmarking demonstrates that this cost is application dependent. The Linpack benchmarking shows that the overhead of container virtualization and SDN interconnect is comparable to the native performance.

Vertical Jump Biomechanics Altered With Virtual Overhead Goal.

Virtual environments with real-time feedback can simulate extrinsic goals that mimic real life conditions. The purpose was to compare jump performance and biomechanics with a physical overhead goal (POG) and with a virtual overhead goal (VOG). Fourteen female subjects participated (age: 18.8 ± 1.1 years, height: 163.2 ± 8.1 cm, weight 63.0 ± 7.9 kg). Sagittal plane trunk, hip, and knee biomechanics were calculated during the landing and take-off phases of drop vertical jump with different goal conditions. Repeated-measures ANOVAs determined differences between goal conditions. Vertical jump height displacement was not different during VOG compared with POG. Greater hip extensor moment (P < .001*) and hip angular impulse (P < .004*) were found during VOG compared with POG. Subjects landed more erect with less magnitude of trunk flexion (P = .002*) during POG compared with VOG. A virtual target can optimize jump height and promote increased hip moments and trunk flexion. This may be a useful alternative to physical targets to improve performance during certain biomechanical testing, screening, and training conditions.

Reprint of “Robust partial-load experiments with Showstopper”

Concurrent execution of multiple applications leads to varying partial utilization of shared resources. Understanding system behavior in these conditions is essential for making concurrent execution efficient. Unfortunately, anticipating behavior of shared resources at partial utilization in complex systems is difficult, realistic experiments that reproduce and examine such behavior are therefore needed.To facilitate experiments at partial utilization, we present a tool that accurately controls the processor utilization of arbitrary concurrent workloads, either establishing constant partial load or replaying a variable load trace. We validate the ability of the tool to enforce the configured partial utilization on multiple platforms, and use the tool to collect novel information on system behavior at partial utilization levels.In detail, our experiments show how to examine the complex relationship between utilization and throughput, useful for tasks such as performance debugging or system dimensioning, and we show this relationship for the DaCapo benchmarks. Further, we show that CPU pinning (a technique used to improve workload isolation) can benefit from dynamic response to system utilization, improving system efficiency with partial utilization. Finally, we show that the overhead of virtualization also changes with partial utilization and CPU allocation.

CloudEVBench - Virtualization Technology Efficiency Testing Tool for the Distributed Infrastructures

The virtualized systems are today a very popular topic and their using plays the great role in current datacenters. The virtualization efficiency is a very important aspect in the real system deployment. Some studies have been published about this topic [1], mainly are based on various benchmarking techniques and are integrated into the specialized testing tools. Such a benchmark tool, which is able to simulate the behavior of a real computing system under the stress of different virtualization configurations, can e.g. well answer the question how many virtual machines could be simultaneously executed on it and how big the virtualization overhead is [2]. We developed and applied a new benchmark tool, which is able to measure the virtualization efficiency and overhead in the virtualization environment.

Case of ARM emulation optimization for offloading mechanisms in Mobile Cloud Computing

The Mobile Cloud Computing (MCC) paradigm depends on efficient offloading of computation from the resource constrained mobile device to the resource rich cloud server. The computational offloading is assisted by system virtualization, application virtualization, and process state migration. However, system and application virtualization techniques force unnecessary overhead on applications that require offloading to the cloud and applications that do not. Moreover, smartphones and cloud data centers are based on heterogeneous processor architectures, such as, ARM and x86. As a result, process migrated from a smartphone needs translation or emulation on the cloud server. Therefore, instruction emulation is a necessary criterion for a comprehensive MCC framework. In this paper, we evaluate the overhead of the system and application virtualization techniques and emulation frameworks that enable MCC offloading mechanisms. We find that the overhead of system and application virtualization can be as high as 4.51% and 55.18% respectively for the SciMark benchmark. Moreover, ARM to Intel device emulation overhead can be as high as 55.53%. We provide a proof of concept of emulation speedup by utilizing efficient Single Instruction, Multiple Data (SIMD) translations. We conclude that the overhead of virtualization and emulation techniques need to be reduced for efficient MCC offloading frameworks.