Elastic Management of Cluster based Services in the Cloud



Similar documents
Virtual Machine Management with OpenNebula in the RESERVOIR project

Cloud and Virtualization to Support Grid Infrastructures

An Introduction to Virtualization and Cloud Technologies to Support Grid Computing

VM Management for Green Data Centres with the OpenNebula Virtual Infrastructure Engine

OGF25/EGEE User Forum Catania, Italy 2 March 2009

The OpenNebula Standard-based Open -source Toolkit to Build Cloud Infrastructures

OpenNebula An Innovative Open Source Toolkit for Building Cloud Solutions

Enabling Technologies for Cloud Computing

Design and Building of IaaS Clouds

IaaS Cloud Architectures: Virtualized Data Centers to Federated Cloud Infrastructures

OpenNebula Leading Innovation in Cloud Computing Management

Cloud Computing Architecture with OpenNebula HPC Cloud Use Cases

Cloud Computing from an Institutional Perspective

Network Virtualization

OpenNebula Cloud Case Studies

Open Source Cloud Computing Management with OpenNebula

SURFnet Cloud Computing Solutions

Introduction to Cloud Computing

Scheduler in Cloud Computing using Open Source Technologies

Getting Started Hacking on OpenNebula

Deployment of Private, Hybrid & Public Clouds with OpenNebula

The OpenNebula Cloud Platform for Data Center Virtualization

HPC Cloud Computing with OpenNebula

An Experimental Study of Load Balancing of OpenNebula Open-Source Cloud Computing Platform

Cloud Computing with Red Hat Solutions. Sivaram Shunmugam Red Hat Asia Pacific Pte Ltd.

How To Compare Cloud Computing To Cloud Platforms And Cloud Computing

Building Clouds with OpenNebula 3.4

Eucalyptus LSS: Load-Based Scheduling on Virtual Servers Using Eucalyptus Private Cloud

Challenges in Hybrid and Federated Cloud Computing

Sistemi Operativi e Reti. Cloud Computing

How To Create A Cloud Based System For Aaas (Networking)

CLOUD COMPUTING. Virtual Machines Provisioning and Migration Services Mohamed El-Refaey

An Introduction to Private Cloud

2) Xen Hypervisor 3) UEC

Infrastructure as a Service (IaaS)

Manjrasoft Market Oriented Cloud Computing Platform

CLEVER: a CLoud-Enabled Virtual EnviRonment

OpenStack Introduction. November 4, 2015

Using SouthBound APIs to build an SDN Solution. Dan Mihai Dumitriu Midokura Feb 5 th, 2014

Hybrid Cloud: Overview of Intercloud Fabric. Sutapa Bansal Sr. Product Manager Cloud and Virtualization Group

VIRTUAL RESOURCE MANAGEMENT FOR DATA INTENSIVE APPLICATIONS IN CLOUD INFRASTRUCTURES

An Open Source Solution for Virtual Infrastructure Management in Private and Hybrid Clouds

THE EUCALYPTUS OPEN-SOURCE PRIVATE CLOUD

MAny-Task Computing (MTC) paradigm [1] embraces

Deploying Business Virtual Appliances on Open Source Cloud Computing

Group Based Load Balancing Algorithm in Cloud Computing Virtualization

Key Research Challenges in Cloud Computing

Private Cloud Database Consolidation with Exadata. Nitin Vengurlekar Technical Director/Cloud Evangelist

Auto-Scaling Model for Cloud Computing System

Solution for private cloud computing

Lecture 02b Cloud Computing II

Building Clouds with OpenNebula 2.2 and StratusLab

Figure 1. The cloud scales: Amazon EC2 growth [2].

Plug-and-play Virtual Appliance Clusters Running Hadoop. Dr. Renato Figueiredo ACIS Lab - University of Florida

OVERLAYING VIRTUALIZED LAYER 2 NETWORKS OVER LAYER 3 NETWORKS

OpenNebula Open Souce Solution for DC Virtualization. C12G Labs. Online Webinar

Vyatta Network OS for Network Virtualization

FREE AND OPEN SOURCE SOFTWARE FOR CLOUD COMPUTING SERENA SPINOSO FULVIO VALENZA

Network performance in virtual infrastructures

Introduction to OpenStack

OpenNebula Cloud Platform for Data Center Virtualization

Eucalyptus: An Open-source Infrastructure for Cloud Computing. Rich Wolski Eucalyptus Systems Inc.

Eucalyptus: An Open-source Infrastructure for Cloud Computing. Rich Wolski Eucalyptus Systems Inc.

SLA Driven Load Balancing For Web Applications in Cloud Computing Environment

OpenNebula Latest Innovations in Private Cloud Computing

This presentation provides an overview of the architecture of the IBM Workload Deployer product.

T Mobile Cloud Computing Private Cloud & Assignment

Manjrasoft Market Oriented Cloud Computing Platform

OpenNebula Open Souce Solution for DC Virtualization

What Is It? Business Architecture Research Challenges Bibliography. Cloud Computing. Research Challenges Overview. Carlos Eduardo Moreira dos Santos

Sean Bennett. Cloud Platforms & Networking Group

Efficient Cloud Management for Parallel Data Processing In Private Cloud

Who s Endian?

Virtualization Technologies and Blackboard: The Future of Blackboard Software on Multi-Core Technologies

Provisioning and Resource Management at Large Scale (Kadeploy and OAR)

An Architecture Model of Sensor Information System Based on Cloud Computing

This presentation covers virtual application shared services supplied with IBM Workload Deployer version 3.1.

VDI Gen2: Quick Reference Guide Top 5 Questions to Ask a VDI Vendor Before Making a Purchase

OpenNebula Open Souce Solution for DC Virtualization

Today: Data Centers & Cloud Computing" Data Centers"

PES. Batch virtualization and Cloud computing. Part 1: Batch virtualization. Batch virtualization and Cloud computing

CumuLogic Load Balancer Overview Guide. March CumuLogic Load Balancer Overview Guide 1

Transcription:

First Workshop on Automated Control for Datacenters and Clouds (ACDC09) June 19th, Barcelona, Spain Elastic Management of Cluster based Services in the Cloud Rafael Moreno Vozmediano, Ruben S. Montero, Ignacio M. Llorente Distributed Systems Architecture Research Group Universidad Complutense de Madrid 1/23

Contents 2/19 Introduction Elastic management and implementation of clustered services Experimental results Conclusions Demo

Introduction Goals Deployment of clustered services on top of a virtualized infrastructure consisting of: Local (in house) virtual infrastructure Cloud resource provider (Amazon EC2) VM manager (the OpenNebula engine) Service Users Virtual service infrastructure Service Layer Capacity Manager OpenNebula Core OpenNebula XEN / KVM Drivers EC2 Driver Cloud Provider (Amazon EC2) Infrastructure Layer Local Infrastructure 3/19

Introduction Motivation of the work The proposed architecture offers several benefits The virtualization layer allows to separate the infrastructure management (VM provisioning) from the service management Elastic provisioning of virtual resources on demand, to adapt the infrastructure to the service requirements. Fully transparent for the service itself, and independent of the type of service The integration with the cloud provides additional capacity to the virtual infrastructure 4/19

Introduction The VM Manager (OpenNebula) Capacity Manager OpenNebula OpenNebula Core XEN / KVM Drivers EC2 Driver http://opennebula.org OpenNebula Core: manages the life cycle of a VM by performing basic VM operations (e.g. deployment, monitoring, or termination) Capacity Manager (scheduler): Adjusts the placement of VMs based on a set of configurable provisioning policies Virtualizer Access Drivers: Pluggable drivers that provide an abstraction of the underlying virtualization layer (Xen, KVM, EC2), by exposing the basic functionality of the hypervisor or cloud interface 5/19

Contents 6/19 Introduction Elastic management and implementation of clustered services Experimental results Conclusions Demo

Elastic manag. and implement. of clustered services Benefits Elastic cluster capacity The capacity of the cluster can be modified by deploying (or shutting down) virtual cluster nodes on demand Cluster partitioning The available physical and cloud resources can be used to deploy VMs bound to different cluster based services This approach isolates the different workloads and the performance assigned to each cluster Heterogeneous configurations The VMs of a service could have multiple software configurations The different service images can be maintained with a minimal operational cost ( install once deploy many approach) 7/19

Implementation of clustered services in the cloud Implementation of a computing cluster Batch job submission system Sun Grid Engine (SGE) Cluster components Front end node: SGE master Back end nodes: SGE workers Job submission requests Front end server node (SGE master host) Service LAN Virtualized back end nodes (SGE worker nodes) OpenNebula Hypervisor (XEN) Hypervisor (XEN) Amazon EC2 Physical resource pool 8/19

Implementation of clustered services in the cloud Implementation of a web server cluster Based on NGINX software Cluster components Front end node: NGINX reverse proxy Back end nodes: NGNIX web servers HTTP Client HTTP Client Internet HTTP Client Front end server node (NGINX reverse proxy) Service LAN Virtualized back end nodes (NGINX web servers) OpenNebula Hypervisor (XEN) Hypervisor (XEN) Amazon EC2 Physical resource pool 9/19

Implementation of clustered services in the cloud The network infrastructure (i) The service LAN Every virtual node (both local and Amazon EC2 nodes) communicates with the front end through the service LAN (using private addresses) The local (in house) virtual nodes Are directly attached to the service LAN by means of a virtual bridge configured in every physical host The remote (Amazon EC2) virtual nodes Are connected to the service LAN by means of a virtual private network (VPN) tunnel The VPN tunnel is implemented with the OpenVPN software 10/19

Implementation of clustered services in the cloud The network infrastructure (ii) Amazon EC2 OpenVPN clients Back end Node Back end Node Back end Node OpenVPN Tunnels Internet Connection Bridge Front end Node + OpenVPN Server Physical Host 0 Bridge LAN (Gigabit Ethernet) Bridge Bridge Bridge Bridge Back end Node Back end Node Back end Node Back end Node Back end Node Back end Node Physical Host 1 Physical Host 2 Physical Host 3 Physical Host 4 11/19

Contents 12/19 Introduction Elastic management and implementation of clustered services Experimental results Conclusions Demo

Experimental Results 13/19 Computing cluster performance 0L+4R 4L+0R 4L+4R 4L+8R 5,0E 02 4,5E 02 4,0E 02 3,5E 02 Throughput (jobs/sec.) 3,0E 02 2,5E 02 2,0E 02 1,5E 02 1,0E 02 5,0E 03 0,0E+00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Jobs

Experimental Results Web server cluster performance Throughput (requests/s) 300 HTTP requests (static file 1KB) 140 120 100 80 60 40 20 0 4L 8L 4L+4R 4L+8R 4L+16R Cluster configuration Throughput (requests/s) 120 100 80 60 40 20 0 300 HTTP requests (static file 10KB) 4L 8L 4L+4R 4L+8R 4L+16R Cluster configuration Throughput (requests/s) 120 100 80 60 40 20 300 HTTP requests (static file 100KB) Throughput (requests/s) 120 100 80 60 40 20 300 HTTP requests (static file 1000KB) 0 4L 8L 4L+4R 4L+8R 4L+16R 0 4L 8L 4L+4R 4L+8R 4L+16R Cluster configuration Cluster configuration 14/19

Contents 15/19 Introduction Elastic management and implementation of clustered services Experimental results Conclusions Demo

Conclusions We have analyzed the deployment of two cluster based services on top of a virtualized infrastructure, with the capacity of integrating external cloud resources. Computing cluster Web server cluster This approach separates the resource provisioning from the service management, and provides important benefits: Elastic capacity to adapt the cluster to its dynamic workload Cluster partitioning to isolate it from other running services Heterogeneous configurations tailored for each application class Performance results show A sustained performance increment when adding a growing number of cloud nodes to the cluster That proves the feasibility of the proposed architecture and provisioning model, and its capacity to support service elasticity. 16/19

Contents 17/19 Introduction Elastic management and implementation of clustered services Experimental results Conclusions Demo

Elastic Management of Cluster based Services in the Cloud 18/19 THANK YOU FOR YOUR ATTENTION QUESTIONS?