Hypervisor and Neighbors’ Noise: Performance Degradation in Virtualized Environments

Users expect isolated performance from rented virtual machines (VMs) in an infrastructure as a service (IaaS) cloud environment. However, this is not happening in todays’ systems because basically VMs are running in a shared environment. In this paper, we study performance degradation in a virtualized environment similar to IaaS clouds using Parsec 2.1 benchmarks. We consider slowdowns caused by hypervisor—hypervisor's noise—as well as co-located VMs—neighbors’ noise. Previous researches did not consider multi-virtual CPU (vCPU) VMs in an overcommitted environments similar to IaaS clouds. Our target system consists of multiple multi-processor VMs running on a commodity chip-multiprocessor by a hypervisor. This configuration is widespread in todays’ IaaS clouds like Amazon EC2. We find that performance degradation in a virtualized environment could be up to <inline-formula> <tex-math notation="LaTeX">$16\times$</tex-math><alternatives><inline-graphic xlink:href="nikounia-ieq1-2464811.gif"/> </alternatives></inline-formula> which is far more than previous findings. Beside shared resources of memory sub-system, blindness of hypervisor's scheduler have large impact on the slowdown and this is contrary to recent researches that mostly blame last-level cache (LLC) contention for performance degradation. After investigating the causes of performance degradation, we provide some ideas that motivate researchers to reduce performance degradation through hardware and software techniques. We also mention some hints that help organizations to see if their applications are ready for the cloud.

[1]  Xiaorong Li,et al.  Evaluating hardware-assisted virtualization for deploying HPC-as-a-service , 2013, VTDC '13.

[2]  Jie Liu,et al.  Cuanta: quantifying effects of shared on-chip resource interference for consolidated virtual machines , 2011, SoCC.

[3]  Kevin Skadron,et al.  Bubble-up: Increasing utilization in modern warehouse scale computers via sensible co-locations , 2011, 2011 44th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO).

[4]  Joshua LeVasseur,et al.  Towards Scalable Multiprocessor Virtual Machines , 2004, Virtual Machine Research and Technology Symposium.

[5]  Johan Tordsson,et al.  An Autonomic Approach to Risk-Aware Data Center Overbooking , 2014, IEEE Transactions on Cloud Computing.

[6]  John K. Ousterhout.  Scheduling Techniques for Concurrebt Systems. , 1982, ICDCS 1982.

[7]  G. Bruce Berriman,et al.  Scientific workflow applications on Amazon EC2 , 2010, 2009 5th IEEE International Conference on E-Science Workshops.

[8]  Xiao Zhang,et al.  CPI2: CPU performance isolation for shared compute clusters , 2013, EuroSys '13.

[9]  Peter Kilpatrick,et al.  IO performance prediction in consolidated virtualized environments , 2011, ICPE '11.

[10]  Calton Pu,et al.  An Analysis of Performance Interference Effects in Virtual Environments , 2007, 2007 IEEE International Symposium on Performance Analysis of Systems & Software.

[11]  A. Kivity,et al.  kvm : the Linux Virtual Machine Monitor , 2007 .

[12]  Aman Kansal,et al.  Q-clouds: managing performance interference effects for QoS-aware clouds , 2010, EuroSys '10.

[13]  Samuel Kounev,et al.  A generic approach for architecture-level performance modeling and prediction of virtualized storage systems , 2013, ICPE '13.

[14]  Sally A. McKee,et al.  Understanding PARSEC performance on contemporary CMPs , 2009, 2009 IEEE International Symposium on Workload Characterization (IISWC).

[15]  Rachel Householder,et al.  On Cloud-based Oversubscription , 2014, ArXiv.

[16]  Vijay K. Naik,et al.  Biting Off Safely More Than You Can Chew: Predictive Analytics for Resource Over-Commit in IaaS Cloud , 2012, 2012 IEEE Fifth International Conference on Cloud Computing.

[17]  Koushik Chakraborty,et al.  Supporting Overcommitted Virtual Machines through Hardware Spin Detection , 2012, IEEE Transactions on Parallel and Distributed Systems.

[18]  Lingjia Tang,et al.  Compiling for niceness: mitigating contention for QoS in warehouse scale computers , 2012, CGO '12.

[19]  Samuel Kounev,et al.  I/O Performance Modeling of Virtualized Storage Systems , 2013, 2013 IEEE 21st International Symposium on Modelling, Analysis and Simulation of Computer and Telecommunication Systems.

[20]  Nathan Regola,et al.  Recommendations for Virtualization Technologies in High Performance Computing , 2010, 2010 IEEE Second International Conference on Cloud Computing Technology and Science.

[21]  Yingwei Luo,et al.  A Simple Cache Partitioning Approach in a Virtualized Environment , 2009, 2009 IEEE International Symposium on Parallel and Distributed Processing with Applications.

[22]  Christian Bienia,et al.  PARSEC 2.0: A New Benchmark Suite for Chip-Multiprocessors , 2009 .

[23]  John R. Lange,et al.  Preemptable ticket spinlocks: improving consolidated performance in the cloud , 2013, VEE '13.

[24]  Sanjay Chaudhary,et al.  Application Performance Isolation in Virtualization , 2009, 2009 IEEE International Conference on Cloud Computing.

[25]  Richard McDougall,et al.  Virtualization performance: perspectives and challenges ahead , 2010, OPSR.

[26]  Heeseung Jo,et al.  XHive: Efficient Cooperative Caching for Virtual Machines , 2011, IEEE Transactions on Computers.