Live Migration of Virtualized Edge Networks: Analytical Modeling and Performance Evaluation

Similar documents
Chapter 5 C. Virtual machines

VM Migration, Containers (Lecture 12, cs262a)

CS370 Operating Systems

Live Virtual Machine Migration with Efficient Working Set Prediction

Disaster Recovery-to-the- Cloud Best Practices

CS370 Operating Systems

Disaster Recovery Solution Achieved by EXPRESSCLUSTER

Xen and CloudStack. Ewan Mellor. Director, Engineering, Open-source Cloud Platforms Citrix Systems

University of Alberta. Zhu Pang. Master of Science. Department of Computing Science

Enable Infrastructure Beyond Cloud

Lecture 09: VMs and VCS head in the clouds

Distributed Systems. 31. The Cloud: Infrastructure as a Service Paul Krzyzanowski. Rutgers University. Fall 2013

Converged Platforms and Solutions. Business Update and Portfolio Overview

Logging, Monitoring, and Alerting

PDP : A Flexible and Programmable Data Plane. Massimo Gallo et al.

CLOUD computing is completely reshaping the way users

Survey of ETSI NFV standardization documents BY ABHISHEK GUPTA FRIDAY GROUP MEETING FEBRUARY 26, 2016

Spring 2017 :: CSE 506. Introduction to. Virtual Machines. Nima Honarmand

Symantec System Recovery 2011 Management Solution Technical FAQ

1V Number: 1V0-621 Passing Score: 800 Time Limit: 120 min. 1V0-621

CloudNet: Dynamic Pooling of Cloud Resources by Live WAN Migration of Virtual Machines

K a t h y Meier- H e l l s t e r n, P h D

Network Virtualisation Vision and Strategy_ (based on lesson learned) Telefónica Global CTO

CloudAP: Improving the QoS of Mobile Applications with Efficient VM Migration

HPVM & OpenVMS. Sandeep Ramavana OpenVMS Engineering Sep Germany Technical Update Days 2009

Towards Converged SmartNIC Architecture for Bare Metal & Public Clouds. Layong (Larry) Luo, Tencent TEG August 8, 2018

Operating Systems 4/27/2015

Performance Considerations of Network Functions Virtualization using Containers

Communication System Design Projects. Communication System Design:

Upgrading Your System a Telco User Perspective. Ulrich Kleber San Francisco November 2015

The Realities of Virtualization

1V0-621.testking. 1V VMware Certified Associate 6 - Data Center Virtualization Fundamentals Exam

vrealize Operations Management Pack for NSX for vsphere 3.0

Live Migration of Virtual Machines

PeopleSoft on Oracle Cloud Platform: Built for Enterprise. Copyright 2017, Oracle and/or its affiliates. All rights reserved.

Understanding Cloud Migration. Ruth Wilson, Data Center Services Executive

BUILDING A PATH TO MODERN DATACENTER OPERATIONS. Virtualize faster with Red Hat Virtualization Suite

Data Path acceleration techniques in a NFV world

Ethernet Fabrics- the logical step to Software Defined Networking (SDN) Frank Koelmel, Brocade

Infrastructure Provisioning with System Center Virtual Machine Manager

Empowering SDN SOFTWARE-BASED NETWORKING & SECURITY FROM VYATTA. Bruno Barba Systems Engineer Mexico & CACE

Parallels Virtuozzo Containers

ITRI Cloud OS: An End-to-End OpenStack Solution

How Symantec Backup solution helps you to recover from disasters?

Chapter 3 Virtualization Model for Cloud Computing Environment

On-Premises Cloud Platform. Bringing the public cloud, on-premises

Cloud and Datacenter Networking

IBM Bluemix compute capabilities IBM Corporation

How it can help your organisation

Consolidated Disaster Recovery. Paul Kangro Applied Technology Strategiest

Annual Public Safety PSAP Survey results

CHAPTER 16 - VIRTUAL MACHINES

NEC Express5800 R320f Fault Tolerant Servers & NEC ExpressCluster Software

Migration Strategies from vsphere to Linux and OpenStack via a shared virtualized network

Virtualization with Arcserve Unified Data Protection

Build Cloud like Rackspace with OpenStack Ansible

BUILDING RESILIENCE in PRODUCTION MIGRATIONS. Sangeeta Handa Billing Infrastructure Engineering

OpenStack Networking: Where to Next?

Disclaimer This presentation may contain product features that are currently under development. This overview of new technology represents no commitme

Bacula Systems Virtual Machine Performance Backup Suite

Distributed Systems COMP 212. Lecture 18 Othon Michail

Securing your Virtualized Datacenter. Charu Chaubal Senior Architect, Technical Marketing 6 November, 2008

Video-Aware Networking: Automating Networks and Applications to Simplify the Future of Video

Virtualization and the Metrics of Performance & Capacity Management

VMware Site Recovery Technical Overview First Published On: Last Updated On:

CSC 5930/9010 Cloud S & P: Virtualization

Empirical Evaluation of Latency-Sensitive Application Performance in the Cloud

Vshadow: Promoting Physical Servers into Virtualization World

Scalable Cloud Management with Management Objectives

Herding virtual workstations at Google

Welcome to the. Migrating SQL Server Databases to Azure

vrealize Operations Management Pack for NSX for vsphere 3.5.0

The OnApp Cloud Platform

Availability & Resource

Transforming Data Protection with HPE: A Unified Backup and Recovery June 16, Copyright 2016 Vivit Worldwide

Nested Virtualization and Server Consolidation

The threat landscape is constantly

TALK THUNDER SOFTWARE FOR BARE METAL HIGH-PERFORMANCE SOFTWARE FOR THE MODERN DATA CENTER WITH A10 DATASHEET YOUR CHOICE OF HARDWARE

Docker and Splunk Development

Paperspace. Architecture Overview. 20 Jay St. Suite 312 Brooklyn, NY Technical Whitepaper

Virtualization And High Availability. Howard Chow Microsoft MVP

CXS Citrix XenServer 6.0 Administration

Thomas Lin, Naif Tarafdar, Byungchul Park, Paul Chow, and Alberto Leon-Garcia

COMPUTER ARCHITECTURE. Virtualization and Memory Hierarchy

Paul Hodge Virtualization Solutions: Improving Efficiency, Availability and Performance

An overview of virtual machine architecture

1 Energy Efficient Protocols in Self-Aware Networks

Symantec System Recovery 2013 Management Solution FAQ

Cisco Cloud Application Centric Infrastructure

High-reliability, High-availability Cluster System Supporting Cloud Environment

Enterprise Cloud Computing. Eddie Toh Platform Marketing Manager, APAC Data Centre Group Cisco Summit 2010, Kuala Lumpur

Course Review. Hui Lu

DC: Le Converged Infrastructure per Software Defined e Cloud Cisco NetApp - Softway. Luigi MARCOCCHIA SOFTWAY

Virtualization. Michael Tsai 2018/4/16

Follow the Sun through the Clouds: Application Migration for Geographically Shifting Workloads

SDN+NFV Next Steps in the Journey

Better Security with Virtual Machines

Oracle IaaS, a modern felhő infrastruktúra

Multiprocessor Scheduling. Multiprocessor Scheduling

Deployment Patterns using Docker and Chef

Transcription:

Live Migration of Virtualized Edge Networks: Analytical Modeling and Performance Evaluation Walter Cerroni, Franco Callegati DEI University of Bologna, Italy

Outline Motivations Virtualized edge networks Live migration of virtual machines (VMs) Multiple VM migration model sequential migration parallel migration Numerical results Conclusion 2

Motivations Traditional IP network architecture is showing all its limitations, especially at network edge heterogeneity of today's application service requirements complexity of cross-layer network administration and management multiplicity of L4 L7 network functions executed by many closed and specialized middle-boxes Emerging technologies to foster a paradigm shift SDN to open network devices, separate bare-metal from network intelligence and ease new service deployment NFV to transform middle-boxes into software apps running on standard HW and simplify network administration and management Bring the advantages of cloud to edge networks 3

Reference Scenario User Data Centers at the Edge Mobile User Virtualized Edge Network Core Network User User 4

Virtualized Edge Networks Media Server Web Server User Access Router Firewall NAT Switch Edge Router VM VM VM VM VM Virtual Bridges/Switches SDN NFV User Hypervisor Kernel Standard HW Edge Router 5

Preliminary Test Setup A. Manzalini et al., Clouds of Virtual Machines in Edge Networks, IEEE Com. Mag., July 2013.

Performance with Off-the-Shelf Technology A. Manzalini et al., Clouds of Virtual Machines in Edge Networks, IEEE Com. Mag., July 2013.

Live Migration of Virtual Machines Service virtualization is a widely used technique for data center administration and maintenance Advantages of OS virtualization (Virtual Machines) quick deployment of new service instances effective load balancing and server consolidation easy service replication and migration mobility easy backup and restore procedures Live migration of VMs current state of VM s kernel and running processes is maintained SDN can help maintaining also the network state no need to wait for long shut-down and restart phases no risk of inconsistencies due to duplicate running VM instances clients do not need to disconnect and reconnect DC providers and customers work on fully separated domains 8

Live Migration of Virtual Machines Focus on memory migration storage migration (if needed) through NAS syncronization network state migration through SDN Two approaches pre-copy: push most of the memory pages to destination host before stopping VM at source host post-copy: pull most of the memory pages from source host after resuming VM at destination host We assume the pre-copy approach (adoped by Xen, KVM, VirtualBox, etc.) iterative push phase: memory pages modified in a given round are sent again in the next round, until total size of dirty pages is below a given threshold or a maximum number of iteration is reached stop-and-copy phase: VM is suspended at source host and the remaining dirty pages are copied to destination resume phase: VM is resumed at destination with consistent memory and network state 9

Performance Metrics for VM Live Migration Downtime ( ): amount of time the VM is suspended measures the user s perceived quality Total Migration Time ( ): amount of time needed to copy the whole memory measures the impact of the migration process on both communication infrastructure and computing resource utilization iterative push phase stop-and-copy phase resume phase time copied memory pages dirtied memory pages 10

Multiple VM Live Migration Model number of mutually dependent VMs in the set to be migrated memory size of VM, page dirtying rate of VM memory page size of VM bit rate used to transfer VM number of iterations needed to migrate VM amount of dirty memory of VM to be copied in round, duration of round for VM 11

Simplified Model all VMs in the set have the same amount of memory all VMs in the set show the same fixed page dirtying rate all VMs in the set have the same memory page size the bit rate dedicated to the migration of VM is fixed condition for pre-copy algorithm to be sustainable dirty memory size threshold max no. of iterations total migration time of VM 12

Performance of Multiple VM Live Migration Correlations among VMs require new definitions of performance metrics for the whole set of VMs Total Migration Time starts when the first VM begins the push phase ends when the last VM ends the stop-and-copy phase Downtime starts when the first VM begins the stop-and-copy phase ends when the last VM ends the resume phase Both depend on order of migration migration scheduling strategy amount of bandwidth used to perform VM migration We analize two simple strategies sequential migration parallel migration 13

Sequential vs. Parallel VM Migration Sequential Migration of one VM at a time at full channel bit rate Parallel Simultaneous migration of all VMs equally sharing the channel bit rate Smaller transfer bit rate but same dirtying rate leads to more iterations in parallel migration than in sequential 14

Sequential vs. Parallel VM Migration Trade-off 15

Results: Role of Page Dirtying Rate 16

Results: Number of Iterations 17

Results: Role of Memory Size 18

Results: Role of Dirty Memory Size Threshold 19

Results: Dimensioning the Channel Bit Rate 20

Results: Role of Number of VMs in the Set 21

Results: Role of Critical Subset Size Critical subset: only m VMs out of M must be running to provide the service 22

Conclusion Need for a paradigm shift, especially for edge networks bring the advantages of cloud infrastructures to edge networks SDN and NFV are the key enablers Virtualized edge networks demonstrated with preliminary tests using VMs need to improve live migration performance Multiple VM live migration model performance depend on migration schedule and resources sequential vs. parallel migration trade off resource usage with user s perceived quality Further study on-going VMs with different memory size different bandwidth allocation stategies trade-off holds in general memory transfer synchronization helps limiting the downtime 23