OSCAR Symposium – Quebec City, Canada – June 2008 Proposal for Modifications to the OSCAR Architecture to Address Challenges in Distributed System Management.

Slides:



Advertisements
Similar presentations
1 OBJECTIVES To generate a web-based system enables to assemble model configurations. to submit these configurations on different.
Advertisements

Network+ Guide to Networks, Fourth Edition Chapter 10 Netware-Based Networking.
Mobile Agents: A Key for Effective Pervasive Computing Roberto Speicys Cardoso & Fabio Kon University of São Paulo - Brazil.
Firefox 2 Feature Proposal: Remote User Profiles TeamOne August 3, 2007 TeamOne August 3, 2007.
Lesson 2: Configuring Servers
11 SERVER CLUSTERING Chapter 6. Chapter 6: SERVER CLUSTERING2 OVERVIEW  List the types of server clusters.  Determine which type of cluster to use for.
Automating Linux Installations at CERN G. Cancio, L. Cons, P. Defert, M. Olive, I. Reguero, C. Rossi IT/PDP, CERN presented by G. Cancio.
System-level Virtualization for HPC: Recent work on Loadable Hypervisor Modules Systems Research Team Computer Science and Mathematics Division Oak Ridge.
Design and Implementation of a Single System Image Operating System for High Performance Computing on Clusters Christine MORIN PARIS project-team, IRISA/INRIA.
Włodzimierz Funika, Filip Szura Automation of decision making for monitoring systems.
Linux Basics CS 302. Outline  What is Unix?  What is Linux?  Virtual Machine.
Client Management. Introduction In a typical organization there are a lot of client machines used for day to day operations Client management is a necessary.
WP6: Grid Authorization Service Review meeting in Berlin, March 8 th 2004 Marcin Adamski Michał Chmielewski Sergiusz Fonrobert Jarek Nabrzyski Tomasz Nowocień.
Linux Operations and Administration
SSI-OSCAR A Single System Image for OSCAR Clusters Geoffroy Vallée INRIA – PARIS project team COSET-1 June 26th, 2004.
 Cloud computing  Workflow  Workflow lifecycle  Workflow design  Workflow tools : xcp, eucalyptus, open nebula.
Computation for Physics 計算物理概論 Introduction to Linux.
A Cloud is a type of parallel and distributed system consisting of a collection of inter- connected and virtualized computers that are dynamically provisioned.
High Performance Computing Cluster OSCAR Team Member Jin Wei, Pengfei Xuan CPSC 424/624 Project ( 2011 Spring ) Instructor Dr. Grossman.
Visual Linker Final presentation.
Eric Keller, Evan Green Princeton University PRESTO /22/08 Virtualizing the Data Plane Through Source Code Merging.
Presented by: Sanketh Beerabbi University of Central Florida COP Cloud Computing.
Virtualization for Adaptability Project Presentation CS848 Fall 2006 Umar Farooq Minhas 29 Nov 2006 David R. Cheriton School of Computer Science University.
Mr C Johnston ICT Teacher
Update on Database Issues Peter Chochula DCS Workshop, June 21, 2004 Colmar.
Client – Server Application Can you create a client server application: The server will be running as a service: does not have a GUI The server will run.
Loosely Coupled Parallelism: Clusters. Context We have studied older archictures for loosely coupled parallelism, such as mesh’s, hypercubes etc, which.
Presented by System-Level Virtualization & OSCAR-V Stephen L. Scott Thomas Naughton Geoffroy Vallée Computer Science Research Group Computer Science and.
DISTRIBUTED SYSTEMS Principles and Paradigms Second Edition ANDREW S
Java Portals and Portlets Submitted By: Rashi Chopra CIS 764 Fall 2007 Rashi Chopra.
Beowulf Software. Monitoring and Administration Beowulf Watch 
SONIC-3: Creating Large Scale Installations & Deployments Andrew S. Neumann Principal Engineer, Progress Sonic.
GAAIN Virtual Appliances: Virtual Machine Technology for Scientific Data Analysis Arihant Patawari USC Stevens Neuroimaging and Informatics Institute July.
ABone Architecture and Operation ABCd — ABone Control Daemon Server for remote EE management On-demand EE initiation and termination Automatic EE restart.
Enabling Grids for E-sciencE EGEE and gLite are registered trademarks Usage of virtualization in gLite certification Andreas Unterkircher.
Experiment Management System CSE 423 Aaron Kloc Jordan Harstad Robert Sorensen Robert Trevino Nicolas Tjioe Status Report Presentation Industry Mentor:
11 CLUSTERING AND AVAILABILITY Chapter 11. Chapter 11: CLUSTERING AND AVAILABILITY2 OVERVIEW  Describe the clustering capabilities of Microsoft Windows.
EXPOSING OVS STATISTICS FOR Q UANTUM USERS Tomer Shani Advanced Topics in Storage Systems Spring 2013.
Architecture View Models A model is a complete, simplified description of a system from a particular perspective or viewpoint. There is no single view.
Linux Operations and Administration
IBM Express Runtime Quick Start Workshop © 2007 IBM Corporation Deploying a Solution.
Computing Facilities CERN IT Department CH-1211 Geneva 23 Switzerland t CF CC Monitoring I.Fedorko on behalf of CF/ASI 18/02/2011 Overview.
Planning Server Deployments Chapter 1. Server Deployment When planning a server deployment for a large enterprise network, the operating system edition.
6MoN plus geographically distributed dual stack network monitoring #TNC16 | #IIT-CNR | #6MoN Speaker: Abraham Gebrehiwot.
RuggedPOD O/S Deployment strategy. Disclaimers The content of this presentation is released under GPL v2 license en Creative Common Attribution-ShareAlike.
Towards a High Performance Extensible Grid Architecture Klaus Krauter Muthucumaru Maheswaran {krauter,
XNAT 1.7: Getting Started 6 June, Introduction In this presentation we’ll discuss:  Features and functions in XNAT 1.7  Requirements  Installing.
BY: SALMAN 1.
Appium Studio Appium testing made easy at any scale.
Agenda:- DevOps Tools Chef Jenkins Puppet Apache Ant Apache Maven Logstash Docker New Relic Gradle Git.
Use of HLT farm and Clouds in ALICE
Administration Tools Cluster.exe is a command line tool that you can use for scripting or remote administration through slow WAN links. Cluadmin.exe is.
Microsoft® System Center Virtual Machine Manager 2008
BY: SALMAN.
System-Level Virtualization & OSCAR-V
Virtualization Review and Discussion
File System Implementation
Shared Services with Spotfire
Chapter 2: Operating-System Structures
GWE Core Grid Wizard Enterprise (
Consulting Services JobScheduler Architecture Decision Template
Operating Systems Overview
Dev Test on Windows Azure Solution in a Box
20409A 7: Installing and Configuring System Center 2012 R2 Virtual Machine Manager Module 7 Installing and Configuring System Center 2012 R2 Virtual.
Intro. To Operating Systems
How to Improve Releasing Efficiency via i18N/L10n Test Automation.
Open Automation Software
A Virtual Machine Monitor for Utilizing Non-dedicated Clusters
PyWBEM Python WBEM Client: Overview #2
GENEDI EUROPEAN COMMISSION - EUROSTAT GENERIC EDI TOOLBOX
Presentation transcript:

OSCAR Symposium – Quebec City, Canada – June 2008 Proposal for Modifications to the OSCAR Architecture to Address Challenges in Distributed System Management Geoffroy Vallée (ORNL) ‏ Thomas Naughton (ORNL) ‏ Stephen L. Scott (ORNL) ‏

OSCAR 2008 – Quebec City, Canada - 2 Context  OSCAR initially has been designed for Beowulf clusters  a single head-node hosting all distributed services  a set of compute nodes  GUI based (a CLI is under development but still unstable)‏  OSCAR use cases by the System Research ORNL  Deployment of system-level virtualization prototypes  Deployment of fault tolerance prototypes  Different Linux distributions  Remote connection  Use to have “UNIX-like tools”

OSCAR 2008 – Quebec City, Canada - 3 Current OSCAR Architecture

OSCAR 2008 – Quebec City, Canada - 4 Advantages & Limitations of the Current Implementation  Advantages  Extensible (support of many Linux distributions)‏  Wizard-based GUI  Limitations  No possible to extend system partitioning: e.g., I/O versus login nodes  Difficult to extend the current GUI: no separation of the GUI code from the “core code”  Difficult to support system-level virtualization  No fully-featured CLI, no UNIX-like UI  No integrated solution for system monitoring (basis for fault tolerance and resiliency)‏

OSCAR 2008 – Quebec City, Canada - 5 System Management Challenges  What are our current challenges?  Remote management  User interfaces  Unix-like tools  Update management  System partitioning  System monitoring  Simplicity (KISS – Keep It Stupid Simple)‏  Goal of this study  extend the current OSCAR architecture in order to address these challenges  be able to use OSCAR for our research

OSCAR 2008 – Quebec City, Canada - 6 Proposed New OSCAR Architecture

OSCAR 2008 – Quebec City, Canada - 7 New OSCAR Core Algorithm  The user setup a new partition  Then, OSCAR performs automatically the following tasks  creates a basic golden image for the partition if no image is available  installs OPKG(s) into the basic golden image  execute post-installation scripts for the installed OPKG(s)‏  assign compute nodes to the image  wait for the compute nodes to contact the server for deployment  when all compute nodes are deployed, execute the configuration scripts  Compared to the current algorithm, mush more details are hidden to the users

OSCAR 2008 – Quebec City, Canada - 8 OPM/Nest  Not a new idea  Implement the smart element of OSCAR  takes the partition configurations  figure out what to do  perform the different tasks, hiding details to users as much as possible  Key idea is the separation of the system configuration and the actual deployment/management tasks  the system configuration is “static” (does not change all the time)‏  configuration saved in ODA  the configuration saved in ODA is used as reference

OSCAR 2008 – Quebec City, Canada - 9 ODA Extension  Central role  store partitions configuration  used by “smart” tools to perform administration tasks  However, ODA was only based on a real database  Difficult to extend  Make new develops more difficult (database modifications take a lot of time)‏  Flat file ODA mode  Still possible to use a real database  Also possible to use flat files (in /etc/oscar/)‏  Simple to extend and modify  Perfect for development

OSCAR 2008 – Quebec City, Canada - 10 ODA Flat Files – Examples ‘-- clusters ‘-- oscar ‘-- partition1 |-- node1 | ‘-- node1.conf |-- node2 | |-- node2.conf | ‘-- vm.conf ‘-- partition1.conf

OSCAR 2008 – Quebec City, Canada - 11 New User Interfaces  Because we try to hide as many details as possible from users, the current CLI was difficult to adapt  New UI requirements  Provide both a CLI & a GUI, based on the same code  CLI requirements  UNIX-like (or POSIX-like)‏  Minimize the number of commands  GUI requirements  possibility to “connect” to remote clusters for a remote management  GUI is running locally  OSCAR commands are executed remotely  use new GUI technologies

OSCAR 2008 – Quebec City, Canada - 12 Overview of the New Command Line Interface  Separate OSCAR commands into two parts  configuration  actually management actions  Two new scripts: oscar-config & oscar  similar to normal UNIX scripts  hide details to users

OSCAR 2008 – Quebec City, Canada - 13 Overview of the C++/Qt4 Graphical User Interface

OSCAR 2008 – Quebec City, Canada - 14 Current Status  Configurator has been fixed: possible to parse configuration file from the CLI  XOSCAR (C++/Qt4 GUI) is actively developed (Bob Barbilon, South Dakota University)‏  runs on Debian, Ubuntu, CentOS, and MAC OS  not wizard-based  OSCAR-V integration ongoing  PanYoung – GsoC 2008 – is working on the release of V2M  Full integration planed for OSCAR 5.4  System partitioning  partitioning configuration support available in trunk  deployment still under development  still need to revisit OPKGs

OSCAR 2008 – Quebec City, Canada - 15 Conclusion  Still ongoing work  Ultimate goal  Ease the inclusion of research prototypes  system-level virtualization  fault-tolerance & resiliency  New GUI/CLI  fit needs of some system administrators (“UNIX-like” interface)‏  based on new GUI technologies  allow remote cluster management  Target the inclusion in other tools  OSCAR has a very specific goal  we should ease the integration of OSCAR into other tools that need OSCAR features

OSCAR 2008 – Quebec City, Canada - 16 Contacts Geoffroy Vallée Computer Science Research Group Computer Science and Mathematics Division

OSCAR 2008 – Quebec City, Canada - 17 Questions?