Prioritizing Local Inter-Domain Communication in Xen Sisu Xi, Chong Li, Chenyang Lu, and Christopher Gill Cyber-Physical Systems Laboratory Washington.

Slides:



Advertisements
Similar presentations
Diagnosing Performance Overheads in the Xen Virtual Machine Environment Aravind Menon Willy Zwaenepoel EPFL, Lausanne Jose Renato Santos Yoshio Turner.
Advertisements

Virtual Switching Without a Hypervisor for a More Secure Cloud Xin Jin Princeton University Joint work with Eric Keller(UPenn) and Jennifer Rexford(Princeton)
Virtualization and Cloud Computing. Definition Virtualization is the ability to run multiple operating systems on a single physical system and share the.
XEN AND THE ART OF VIRTUALIZATION Paul Barham, Boris Dragovic, Keir Fraser, Steven Hand, Tim Harris, Alex Ho, Rolf Neugebauer, lan Pratt, Andrew Warfield.
Performance Evaluation of Open Virtual Routers M.Siraj Rathore
VSphere vs. Hyper-V Metron Performance Showdown. Objectives Architecture Available metrics Challenges in virtual environments Test environment and methods.
Project Overview 2014/05/05 1. Current Project “Research on Embedded Hypervisor Scheduler Techniques” ◦ Design an energy-efficient scheduling mechanism.
Realizing Compositional Scheduling through Virtualization Jaewoo Lee, Sisu Xi, Sanjian Chen, Linh T.X. Phan Chris Gill, Insup Lee, Chenyang Lu, Oleg Sokolsky.
Towards High-Availability for IP Telephony using Virtual Machines Devdutt Patnaik, Ashish Bijlani and Vishal K Singh.
Network Implementation for Xen and KVM Class project for E : Network System Design and Implantation 12 Apr 2010 Kangkook Jee (kj2181)
ENFORCING PERFORMANCE ISOLATION ACROSS VIRTUAL MACHINES IN XEN Diwaker Gupta, Ludmila Cherkasova, Rob Gardner, Amin Vahdat Middleware '06 Proceedings of.
Hosted VMM Architecture Advantages: –Installs and runs like an application –Portable – host OS does I/O access –Coexists with applications running on.
Virtualization for Cloud Computing
Xen and the Art of Virtualization. Introduction  Challenges to build virtual machines Performance isolation  Scheduling priority  Memory demand  Network.
Virtualization Technology Prof D M Dhamdhere CSE Department IIT Bombay Moving towards Virtualization… Department of Computer Science and Engineering, IIT.
1 Scheduling I/O in Virtual Machine Monitors© 2008 Diego Ongaro Scheduling I/O in Virtual Machine Monitors Diego Ongaro, Alan L. Cox, and Scott Rixner.
Virtualization Concept. Virtualization  Real: it exists, you can see it.  Transparent: it exists, you cannot see it  Virtual: it does not exist, you.
Networking Virtualization Using FPGAs Russell Tessier, Deepak Unnikrishnan, Dong Yin, and Lixin Gao Reconfigurable Computing Group Department of Electrical.
Dual Stack Virtualization: Consolidating HPC and commodity workloads in the cloud Brian Kocoloski, Jiannan Ouyang, Jack Lange University of Pittsburgh.
Jakub Szefer, Eric Keller, Ruby B. Lee Jennifer Rexford Princeton University CCS October, 2011 報告人:張逸文.
ICT Day Term 4,  Virtualisation is growing in usage.  Current CPU’s are designed to support Virtualisation.  Businesses are looking at virtualisation.
Silberschatz, Galvin and Gagne  Operating System Concepts Chapter 1: Introduction What is an Operating System? Mainframe Systems Desktop Systems.
Virtual Machine Scheduling for Parallel Soft Real-Time Applications
Secure & flexible monitoring of virtual machine University of Mazandran Science & Tecnology By : Esmaill Khanlarpour January.
Xen I/O Overview. Xen is a popular open-source x86 virtual machine monitor – full-virtualization – para-virtualization para-virtualization as a more efficient.
การติดตั้งและทดสอบการทำคลัสเต อร์เสมือนบน Xen, ROCKS, และไท ยกริด Roll Implementation of Virtualization Clusters based on Xen, ROCKS, and ThaiGrid Roll.
Xen I/O Overview.
Improving Network I/O Virtualization for Cloud Computing.
Power and Performance Modeling in a Virtualized Server System M. Pedram and I. Hwang Department of Electrical Engineering Univ. of Southern California.
Virtual Machine Security Systems Presented by Long Song 08/01/2013 Xin Zhao, Kevin Borders, Atul Prakash.
张俊 BTLab Embedded Virtualization Group Outline  Introduction  Performance Analysis  PerformanceTuning Methods.
Penn State CSE “Optimizing Network Virtualization in Xen” Aravind Menon, Alan L. Cox, Willy Zwaenepoel Presented by : Arjun R. Nath.
Politecnico di Torino Dipartimento di Automatica ed Informatica TORSEC Group Performance of Xen’s Secured Virtual Networks Emanuele Cesena Paolo Carlo.
A study of introduction of the virtualization technology into operator consoles T.Ohata, M.Ishii / SPring-8 ICALEPCS 2005, October 10-14, 2005 Geneva,
Challenges towards Elastic Power Management in Internet Data Center.
Our work on virtualization Chen Haogang, Wang Xiaolin {hchen, Institute of Network and Information Systems School of Electrical Engineering.
1 Xen and Co.: Communication-aware CPU Scheduling for Consolidated Xen-based Hosting Platforms Sriram Govindan, Arjun R Nath, Amitayu Das, Bhuvan Urgaonkar,
Xen (Virtual Machine Monitor) Operating systems laboratory Esmail asyabi- April 2015.
CS533 Concepts of Operating Systems Jonathan Walpole.
DESIGNING VM SCHEDULERS FOR EMBEDDED REAL-TIME APPLICATIONS Alejandro Masrur, Thomas Pfeuffer, Martin Geier, Sebastian Drössler and Samarjit Chakraborty.
ECE 720T5 Fall 2012 Cyber-Physical Systems Rodolfo Pellizzoni.
VTurbo: Accelerating Virtual Machine I/O Processing Using Designated Turbo-Sliced Core Embedded Lab. Kim Sewoog Cong Xu, Sahan Gamage, Hui Lu, Ramana Kompella,
Introduction to virtualization
Virtual Machines Created within the Virtualization layer, such as a hypervisor Shares the physical computer's CPU, hard disk, memory, and network interfaces.
Full and Para Virtualization
Technical Reading Report Virtual Power: Coordinated Power Management in Virtualized Enterprise Environment Paper by: Ripal Nathuji & Karsten Schwan from.
Protection of Processes Security and privacy of data is challenging currently. Protecting information – Not limited to hardware. – Depends on innovation.
CSE598c - Virtual Machines - Spring Diagnosing Performance Overheads in the Xen Virtual Machine EnvironmentPage 1 CSE 598c Virtual Machines “Diagnosing.
Inter-domain Socket Communications Supporting High Performance and Full Binary Compatibility on Xen March 5, 2008 Kangho Kim System Software Team Internet.
© 2004 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice Understanding Virtualization Overhead.
Emerging applications in cloud High performance computing E-Commerce Media hosting Web hosting Content delivery... –from Amazon AWS survey 1 Emulated network.
Unit 2 VIRTUALISATION. Unit 2 - Syllabus Basics of Virtualization Types of Virtualization Implementation Levels of Virtualization Virtualization Structures.
1 PERFORMANCE DIFFERENTIATION OF NETWORK I/O in XEN by Kuriakose Mathew ( )‏ under the supervision of Prof. Purushottam Kulkarni and Prof. Varsha.
Is Virtualization ready for End-to-End Application Performance?
Current Generation Hypervisor Type 1 Type 2.
NEWS LAB 薛智文 嵌入式系統暨無線網路實驗室
Container-based Operating System Virtualization: A scalable, High-performance Alternative to Hypervisors Stephen Soltesz, Herbert Potzl, Marc E. Fiuczynski,
The Multikernel: A New OS Architecture for Scalable Multicore Systems
Virtualization overview
XenFS Sharing data in a virtualised environment
Comparison of the Three CPU Schedulers in Xen
Virtualization Layer Virtual Hardware Virtual Networking
Xen Network I/O Performance Analysis and Opportunities for Improvement
RT-Xen: Real-Time Virtualization
Windows Virtual PC / Hyper-V
Introduction to Virtual Machines
Introduction to Virtual Machines
Progress Report 2015/01/28.
Xing Pu21 Ling Liu1 Yiduo Mei31 Sankaran Sivathanu1 Younggyun Koh1
Hypervisor A hypervisor or virtual machine monitor (VMM) is computer software, firmware or hardware that creates and runs virtual machines. A computer.
Presentation transcript:

Prioritizing Local Inter-Domain Communication in Xen Sisu Xi, Chong Li, Chenyang Lu, and Christopher Gill Cyber-Physical Systems Laboratory Washington University in St. Louis IEEE/ACM International Symposium on Quality of Service, 2013

 Multiple computing elements  Cost! Weight! Power!  Communicate via dedicated network or real-time networks  Use fewer computing platforms to integrate independently developed systems via virtualization Motivation 2

 Multiple computing elements  Cost! Weight! Power!  Communicate via dedicated network or real-time networks  Use fewer computing platforms to integrate independently developed systems via virtualization Motivation 3 Physically Isolated Hosts -> Common Computing Platforms Physically Isolated Hosts -> Common Computing Platforms Network Communication -> Local Inter-Domain Communication Network Communication -> Local Inter-Domain Communication Guarantee QoS with Virtualization???

System Model and Contributions  We focus on  Xen as the underlying virtualization software  Single core for each virtual machine on a multi-core platform  Local Inter-Domain Communication (IDC)  No modification to the guest domain besides the Xen patch  Contributions  Real-Time Communication Architecture (RTCA) in Xen  Reduces high priority IDC latency from ms to us in the presence of low priority IDC 4

B Background – Xen Overview 5 A NIC VMM Scheduler Core VCPU netfront Domain 1 VCPU NIC driver softnet_dat a netback Domain 0 VCPU netfront Domain 2 ……

Part I – VMM Scheduler: Limitations  Default credit scheduler  Schedule VCPUs in round-robin order  RT-Xen scheduling framework  Schedule VCPUs by priority  Server based mechanism, each VCPU has (budget, period)  However  If execution time < 0.5 ms, VCPU budget is not consumed  Solution  Dual quanta: ms for scheduling, while us for time accounting 6 “Realizing Compositional Scheduling through Virtualization”, Real-Time and Embedded Technology and Application Symposium (RTAS), 2012 “Realizing Compositional Scheduling through Virtualization”, Real-Time and Embedded Technology and Application Symposium (RTAS), 2012 “RT-Xen: Towards Real-Time Hypervisor Scheduling in Xen”, ACM International Conferences on Embedded Software (EMSOFT), 2011 “RT-Xen: Towards Real-Time Hypervisor Scheduling in Xen”, ACM International Conferences on Embedded Software (EMSOFT), 2011

Part I – VMM Scheduler: Evaluation 7 VMM Scheduler: RT-Xen VS. Credit C 5 C 0C 1 C 3 C 4 Dom 3Dom 4Dom 0Dom 1Dom 2 Linux % CPU sent pkt every 10ms 5,000 data points C 2 Dom 9Dom 10 … When Domain 0 is not busy, the VMM scheduler dominates the IDC performance for higher priority domains

Part I – VMM Scheduler: Enough??? 8 VMM Scheduler C 5 C 0C 1C 2 C 4 C 3 Dom 3 Dom 4 Dom 5Dom 0Dom 1Dom 2 100% CPU … … …

Part II – Domain 0: Background 9 C D A netfront Domain 1 Domain 0 … B netfront Domain 2 … netif TXRX netback netback[0] { rx_action(); tx_action(); } netfront Domain m … netfront Domain n … … … netif softnet_dat a Packets are fetched in a round-robin order Sharing one queue in softnet_data

Part II – Domain 0: RTCA 10 Packets are fetched by priority, up to batch size A netfront Domain 1 Domain 0 … A netfront Domain 2 … netif TXRX netback netback[0] { rx_action(); tx_action(); } … … softnet_dat a B netfront Domain m … netif netfront Domain n … … netif B Queues are separated by priority in softnet_data

Part II – Domain 0: Evaluation Setup 11 VMM Scheduler C 5 C 0C 1C 2 C 4 C 3 Dom 0Dom 1Dom 2 100% CPU Original vs. RTCA Dom 3Dom 4Dom 5 Interference Medium Heavy Light Base … … … sent pkt every 10ms 5,000 data points

Part II – Domain 0: Latency 12 When there is no interference, IDC performance is comparable Original Domain 0 performs poorly in all cases Due to priority inversion within Domain 0 RTCA with batch size 1 performs best We eliminate most of the priority inversions RTCA with larger batch sizes perform worse under IDC interference IDC Latency between Domain 1 and Domain 2 in presence of low priority IDC (us)

Part II – Domain 0: Latency 13 When there is no interference, IDC performance is comparable Original Domain 0 performs poorly in all cases Due to priority inversion within Domain 0 RTCA with batch size 1 performs best we eliminate most of the priority inversions RTCA with larger bath sizes perform worse under IDC interference By reducing priority inversion in Domain 0, RTCA can effectively mitigate impacts of low priority IDC on the latency of high priority IDC IDC Latency between Domain 1 and Domain 2 in presence of low priority IDC (us)

Part II – Domain 0: Throughput 14 A small batch size leads to significant reduction in high priority IDC latency and improved IDC throughput under interfering traffic iPerf Throughput between Dom 1 and Dom 2

Other Approaches and Future Work  Shared Memory Approach [XWAY, XenLoop, Xensocket]  Required modification to guest OS or applications  Traffic Control in Linux [  Applied within one device. Cannot directly be applied on IDC  Future Work  Multi-Core VM scheduling  Network Interface Card (NIC)  Rate control  Co-ordinate with VMM scheduler 15

Conclusion 16 Hardware VMM Scheduler VCPU netfront Domain 1 VCPU softnet_dat a netback Domain 0 VCPU netfront Domain 2  VMM scheduler alone cannot guarantee IDC latency  RTCA: Real-Time Communication Architecture  RTCA + RT-Xen reduces high priority IDC latency from ms to us in the presence of low priority IDC 

Backup Slides 17

Why IDC? Why Xen? 18  Embedded Systems  Integrated Modular Avionics ARINC 653 Standard Honeywell claims that IMA design can save 350 pounds of weight on a narrow-body jet: equivalent to two adults embedded.org/docs/Events/2007/IMA/Slides/ARTIST2_IMA_WindRiv er_Wilson.pdf Full Virtualization based ARINC 653 partition Sanghyun Han, Digital Avionics Systems Conference (DASC), 2011 Full Virtualization based ARINC 653 partition Sanghyun Han, Digital Avionics Systems Conference (DASC), 2011 ARINC 653 Hypervisor VanderLeest S.H., Digital Avionics Systems Conference (DASC), 2010 ARINC 653 Hypervisor VanderLeest S.H., Digital Avionics Systems Conference (DASC), 2010

Latency Matters to Services  Amazon: Revenue decreased by 1% of sales for every 100 ms latency  costs-you-sales-how-to-crush-it.html costs-you-sales-how-to-crush-it.html  Google: slowing down the search results page by 100 ms to 400 ms has a measurable impact on the number of searches per user of -0.2% to - 0.6%   Firefox: 2.2 seconds faster web response increases 15.4% more Firefox install package download. (equals million additional downloads per year)  19

End-to-End Task Performance 20 VMM Scheduler: Credit vs. RT-Xen C 5 C 0C 1C 2 C 4 C 3 Dom 11Dom 12Dom 13 Interference Medium Heavy Light Dom 0Dom 1Dom 2 100% CPU Original vs. RTCA T1(10, 2) T2(20, 2) T1(10, 2) T3(20, 2) T1(10, 2) T4(30, 2) Dom 1 & Dom 2 60% CPU each Dom 3 to Dom 10 10% CPU each 4 pairs bouncing packets Dom 3 Dom 4 Dom 5 Dom 6 Dom 7 Dom 8 Dom 9 Dom 10 Base

End-to-End Task Performance 21 By combining the RT-Xen VMM scheduler and the RTCA Domain 0 kernel, we can deliver end-to-end real- time performance to tasks involving both computation and communication

Backup – Baseline 22

Domain-0 Domain-U (1). XEN Virtual Network 23 socket(AF_INET, SOCKET_DGRAM, 0); socket(AF_INET, SOCKET_STREAM, 0); sendto(…) recvfrom(…) VMM app kernel TCP IP Netback Driver UDP INET TCP IP Netfront Driver UDP INET Transparent Isolation General Migration XPerformance XData Integrity XMulticast

Domain-U (2). XWay, VEE’08 24 VMM XWAY switch TCP IP XWAY protocol NetfrontXWAY driver UDP INET app kernel socket(AF_INET, SOCKET_DGRAM, 0); socket(AF_INET, SOCKET_STREAM, 0); sendto(…) recvfrom(…) Transparent ? Performance Dynamic Create/Destroy Live Migration XConnect Overhead XPatch Guest OS XNo UDP XComplicated

Domain-U (3). XenSocket, Middleware’07 (IBM) 25 VMM app kernel socket(AF_INET, SOCKET_DGRAM, 0); socket(AF_INET, SOCKET_STREAM, 0); socket(AF_XEN, …); sendto(…) recvfrom(…) TCP IP Netfront UDP INET AF_Xen Netfront No Modification to OS/Xen One way Communication Performance XTransparent

Domain-U (4). XenLoop, HPDC’08 (Binghamton) 26 VMM app kernel socket(AF_INET, SOCKET_DGRAM, 0); socket(AF_INET, SOCKET_STREAM, 0); sendto(…) recvfrom(…) TCP IP Netfront UDP INET XenLoop No Modification to OS/Xen Transparent Performance Migration XOverhead XIsolation ? XDynamic teardown ?

A netfront Domain-1 Domain-0 … A netfront Domain-2 … netif … … softnet_data NIC driver multiple kthreads B netfront Domain-m … netif netfront Domain-n … … netif B TXRX netback TXRX netback TXRX netback priority kthreads highest priority