A Survey on Virtual Machine Migration: Challenges, Techniques, and Open Issues

Zhang, Fei; Liu, Guangming; Fu, Xiaoming; Yahyapour, Ramin

doi:10.1109/comst.2018.2794881

Cited by 208 publications

(134 citation statements)

References 186 publications

Supporting

Mentioning

132

Contrasting

Unclassified

Order By: Relevance

“…Typically, the size of the workload is less than 30% of the users' expected maximum, directly translating to the fact that servers run at less than 30% of their maximum capacity [5]. Several works have attempted to tackle the underutilization in FPGA clouds by leveraging the concept of virtual machines to minimize the amount of required resources and turn off the unused resources [18]. In these approaches, FPGA floorplan is split into smaller chunks, called virtual FPGAs, each of which hosts a virtual machine.…”

Section: Introductionmentioning

confidence: 99%

Workload-Aware Opportunistic Energy Efficiency in Multi-FPGA Platforms

Salamat

Khaleghi

Imani

et al. 2019

2019 IEEE/ACM International Conference on Computer-Aided Design (ICCAD)

View full text Add to dashboard Cite

The continuous growth of big data applications with high computational and scalability demands has resulted in increasing popularity of cloud computing. Optimizing the performance and power consumption of cloud resources is therefore crucial to relieve the costs of data centers. In recent years, multi-FPGA platforms have gained traction in data centers as lowcost yet high-performance solutions particularly as acceleration engines, thanks to the high degree of parallelism they provide. Nonetheless, the size of data centers workloads varies during service time, leading to significant underutilization of computing resources while consuming a large amount of power, which turns out as a key factor of data center inefficiency, regardless of the underlying hardware structure. In this paper, we propose an efficient framework to throttle the power consumption of multi-FPGA platforms by dynamically scaling the voltage and hereby frequency during runtime according to prediction of, and adjustment to the workload level, while maintaining the desired Quality of Service (QoS). This is in contrast to, and more efficient than, conventional approaches that merely scale (i.e., power-gate) the computing nodes or frequency. The proposed framework carefully exploits a pre-characterized library of delay-voltage, and power-voltage information of FPGA resources, which we show is indispensable to obtain the efficient operating point due to the different sensitivity of resources w.r.t. voltage scaling, particularly considering multiple power rails residing in these devices. Our evaluations by implementing state-of-the-art deep neural network accelerators revealed that, providing an average power reduction of 4.0×, the proposed framework surpasses the previous works by 33.6% (up to 83%).

show abstract

Section: Introductionmentioning

confidence: 99%

Workload-Aware Opportunistic Energy Efficiency in Multi-FPGA Platforms

Salamat

Khaleghi

Imani

et al. 2019

2019 IEEE/ACM International Conference on Computer-Aided Design (ICCAD)

View full text Add to dashboard Cite

show abstract

“…academy. An updated review on the main proposed solutions, open challenges and research directions is provided, for example, in 5 . However, at the best of the authors' knowledge, the wireless bandwidth management problem afforded by this paper under the 5G FOGRAN scenario of Fig.…”

Section: Related Workmentioning

confidence: 99%

“…(38), (39) and (40) strictly increase (resp., decrease) for increasing values of the dirty rate w (resp., the maximum transport rate R These considerations open the doors to the question regarding the optimized setting of I MAX . Unfortunately, this is still an open question even in the case of state-of-the-art hypervisors, whose typically adopted application-oblivious default setting 9,12 : I MAX = 29 does not guarantee, indeed, the convergence of the iterative pre-copy migration process and does not assure the minimization of the overall consumed energy (see, for example, 5,38 and references therein). However, we anticipate that the carried out tests support the conclusion that, under the following setting of I MAX :…”

Section: Feasibility Conditions and Optimized Tuning Of The Maximum Nmentioning

confidence: 99%

“…In detail, the simulated platform emulates the wireless interconnection between an iPhone7-Plus smartphone and a virtualized Dell Power Edge server that acts as a Fog node. The smartphone is assumed to be equipped with the iOS7 operating system, that already supports MPTCP for enabling its SIRI-based voice recognition applications 5 . The smartphone is assumed to be also equipped with: (i) 1.4 GHz CPU and 128 MB RAN;…”

Section: Performance Testsmentioning

confidence: 99%

See 1 more Smart Citation

Fog-Supported Delay-Constrained Energy-Saving Live Migration of VMs Over MultiPath TCP/IP 5G Connections

Baccarelli¹,

Scarpiniti²,

Momenzadeh³

2018

IEEE Access

View full text Add to dashboard Cite

The incoming era of the Fifth Generation (5G) Fog Computing (FC)-supported Radio Access Networks (shortly, 5G FOGRANs) aims at exploiting the virtualization of the computing/networking resources. The goal is to allow battery-powered wireless devices to augment their limited native resources through the seamless (e.g., live) migration of Virtual Machines (VMs) towards nearby Fog data centers. For this purpose, the bandwidths of the multiple heterogeneous Wireless Network Interface Cards (WNICs) typically equipping the current wireless devices may be used in parallel under the control of the emerging MultiPathTCP (MPTCP) protocol. However, due to the fading and mobility-induced phenomena impairing the underlying RANs, the energy consumptions and migration delays of the current state-of-the-art VM migration techniques may drastically reduce their expected benefits. Motivated by these considerations, in this paper, we analytically characterize, implement in software and numerically test the optimal minimum-energy Settable-Complexity Bandwidth Manager (SCBM) for the live migration of VMs over MPTCP-supported 5G FOGRAN connections. The key features of the proposed SCBM are that: (i) its implementation complexity is settable on-line, in order to attain the targeted energy consumption-vs.-implementation complexity tradeoff; (ii) it minimizes the energy consumed by the migration process under hard QoS constraints on the allowed migration time and downtime;and, (iii) by leveraging a suitably designed adaptive mechanism, it is capable to quickly react to (possibly, unpredicted) fading and/or mobility-induced abrupt changes of the operating conditions without requiring and forecasting support. We numerically test and compare the energy and delay performances of the proposed adaptive SCBM under several 3G/4G/WiFi FOGRAN scenarios by considering a number of synthetic and real-world workloads. The obtained numerical results point out that: (i) the energy savings of the proposed adaptive SCBM over the benchmark one currently implemented by legacy Xen, KWM and VMware hypervisors are typically over 25% and approach 70% when the migrated applications are memory-write intensive; and, (ii) the MPTCP may reduce the energy consumption of the proposed SCBM over legacy SinglePathTCP (SPTCP) more than 50% under strict real-time limits on the allowed migration times.

show abstract

“…Live migration is performed in order to maintain the state of the migrated system same as the original with least interruption to service running inside the VM. Multiple VMs can also be migrated together in case of server failure [7].…”

Section: Introductionmentioning

confidence: 99%

A Distributed API for Live VM Migration in Cloudlets

Singhai¹,

Damania²,

Holmukhe³

et al. 2018

IJCA

View full text Add to dashboard Cite

Cloud computing is an effective technology in handling computation with dynamically scalable resources. With the growth of multimedia applications, mobile applications have become resource-intensive. To provide better connectivity with the cloud, cloudlets have been introduced, which in turns provide low latency and high bandwidth. In this paper, we discuss the live migration of Virtual machines in Cloudlets using VirtualBox as hypervisor. We have proposed and implemented a distributed API Viper particularly for cloudlets, which provides users, interfaces for operating with Virtual Machines. We also test this API in different scenarios, changing RAM, CPU and CPU Stress.

show abstract

A Survey on Virtual Machine Migration: Challenges, Techniques, and Open Issues

Cited by 208 publications

References 186 publications

Workload-Aware Opportunistic Energy Efficiency in Multi-FPGA Platforms

Workload-Aware Opportunistic Energy Efficiency in Multi-FPGA Platforms

Fog-Supported Delay-Constrained Energy-Saving Live Migration of VMs Over MultiPath TCP/IP 5G Connections

A Distributed API for Live VM Migration in Cloudlets

Contact Info

Product

Resources

About