Agreement-based Distributed Resource Management Alain Andrieux Karl Czajkowski.

Slides:



Advertisements
Similar presentations
Distributed Data Processing
Advertisements

Interaction model of grid services in mobile grid environment Ladislav Pesicka University of West Bohemia.
Database Architectures and the Web
Internet Technologies (Grid Computing (OGSA, WSRF) )
High Performance Computing Course Notes Grid Computing.
CLOUD COMPUTING AN OVERVIEW & QUALITY OF SERVICE Hamzeh Khazaei University of Manitoba Department of Computer Science Jan 28, 2010.
SPECIFYING AND MONITORING GUARANTEES IN COMMERCIAL GRIDS THROUGH SLA Sven Graupner Vijay MachirajuAad van Moorsel IEEE/ACM International Symposium on Clustering.
Resource Management of Grid Computing
Universität Dortmund Robotics Research Institute Information Technology Section Grid Metaschedulers An Overview and Up-to-date Solutions Christian.
Introduction and Overview “the grid” – a proposed distributed computing infrastructure for advanced science and engineering. Purpose: grid concept is motivated.
Introduction  What is an Operating System  What Operating Systems Do  How is it filling our life 1-1 Lecture 1.
Milos Kobliha Alejandro Cimadevilla Luis de Alba Parallel Computing Seminar GROUP 12.
Grids and Grid Technologies for Wide-Area Distributed Computing Mark Baker, Rajkumar Buyya and Domenico Laforenza.
Legion Worldwide virtual computer. About Legion Made in University of Virginia Object-based metasystems software project middleware that connects computer.
Operating Systems CS208. What is Operating System? It is a program. It is the first piece of software to run after the system boots. It coordinates the.
DISTRIBUTED COMPUTING
Resource Management in Data-Intensive Systems Bernie Acs, Magda Balazinska, John Ford, Karthik Kambatla, Alex Labrinidis, Carlos Maltzahn, Rami Melhem,
Resource Management Reading: “A Resource Management Architecture for Metacomputing Systems”
Grid Computing 7700 Fall 2005 Lecture 17: Resource Management Gabrielle Allen
Grid Scheduling through Service-Level Agreement Karl Czajkowski The Globus Project
Chapter 9 Elements of Systems Design
Chapter 3: Operating-System Structures System Components Operating System Services System Calls System Programs System Structure Virtual Machines System.
DISTRIBUTED COMPUTING
Frascati, October 9th, Accounting in DataGrid Initial Architecture Albert Werbrouck Frascati, October 9, 2001.
Multicore Resource Management 謝政宏. 2 Outline Background Virtual Private Machines  Spatial Component  Temporal Component  Minimum and Maximum.
1 COMPSCI 110 Operating Systems Who - Introductions How - Policies and Administrative Details Why - Objectives and Expectations What - Our Topic: Operating.
Unit – I CLIENT / SERVER ARCHITECTURE. Unit Structure  Evolution of Client/Server Architecture  Client/Server Model  Characteristics of Client/Server.
PERVASIVE COMPUTING MIDDLEWARE BY SCHIELE, HANDTE, AND BECKER A Presentation by Nancy Shah.
Silberschatz, Galvin and Gagne  Operating System Concepts Chapter 3: Operating-System Structures System Components Operating System Services.
1 4/23/2007 Introduction to Grid computing Sunil Avutu Graduate Student Dept.of Computer Science.
Middleware for Grid Computing and the relationship to Middleware at large ECE 1770 : Middleware Systems By: Sepehr (Sep) Seyedi Date: Thurs. January 23,
Superscheduling and Resource Brokering Sven Groot ( )
NA-MIC National Alliance for Medical Image Computing UCSD: Engineering Core 2 Portal and Grid Infrastructure.
Cracow Grid Workshop ‘06 17 October 2006 Execution Management and SLA Enforcement in Akogrimo Antonios Litke Antonios Litke, Kleopatra Konstanteli, Vassiliki.
What is SAM-Grid? Job Handling Data Handling Monitoring and Information.
Prof S.Ramachandram Dept of CSE,UCE Osmania University
Introduction to Grids By: Fetahi Z. Wuhib [CSD2004-Team19]
Resources Management and Component Placement Presenter:Bo Sheng.
16/11/ Semantic Web Services Language Requirements Presenter: Emilia Cimpian
1 G52IWS: Web Services Chris Greenhalgh. 2 Contents The World Wide Web Web Services example scenario Motivations Basic Operational Model Supporting standards.
International Symposium on Grid Computing (ISGC-07), Taipei - March 26-29, 2007 Of 16 1 A Novel Grid Resource Broker Cum Meta Scheduler - Asvija B System.
AMQP, Message Broker Babu Ram Dawadi. overview Why MOM architecture? Messaging broker like RabbitMQ in brief RabbitMQ AMQP – What is it ?
Globus and PlanetLab Resource Management Solutions Compared M. Ripeanu, M. Bowman, J. Chase, I. Foster, M. Milenkovic Presented by Dionysis Logothetis.
EGEE is a project funded by the European Union under contract IST WS-Based Advance Reservation and Co-allocation Architecture Proposal T.Ferrari,
© 2004 IBM Corporation ICSOC2004 Panel Discussion: Grid Systems: What is needed from web service standards? Jeffrey Frey IBM.
Introduction to Grid Computing and its components.
Aneka Cloud ApplicationPlatform. Introduction Aneka consists of a scalable cloud middleware that can be deployed on top of heterogeneous computing resources.
Providing web services to mobile users: The architecture design of an m-service portal Minder Chen - Dongsong Zhang - Lina Zhou Presented by: Juan M. Cubillos.
GRID ANATOMY Advanced Computing Concepts – Dr. Emmanuel Pilli.
INFSO-RI Enabling Grids for E-sciencE Grid Services for Resource Reservation and Allocation Tiziana Ferrari Istituto Nazionale di.
INTRODUCTION TO GRID & CLOUD COMPUTING U. Jhashuva 1 Asst. Professor Dept. of CSE.
ACGT Architecture and Grid Infrastructure Juliusz Pukacki ‏ EGEE Conference Budapest, 4 October 2007.
9 Systems Analysis and Design in a Changing World, Fifth Edition.
Enabling Grids for E-sciencE Agreement-based Workload and Resource Management Tiziana Ferrari, Elisabetta Ronchieri Mar 30-31, 2006.
Towards a High Performance Extensible Grid Architecture Klaus Krauter Muthucumaru Maheswaran {krauter,
COMPSCI 110 Operating Systems
Grid Optical Burst Switched Networks
Duncan MacMichael & Galen Deal CSS 534 – Autumn 2016
Grid Resource Allocation Agreement Protocol Working Group
Grid Computing.
SOA in Action Chapter 10 B. Ramamurthy 1/16/2019.
Introduction to Grid Technology
Large Scale Distributed Computing
GRUBER: A Grid Resource Usage SLA Broker
Service Oriented Architecture (SOA)
Resource and Service Management on the Grid
The Anatomy and The Physiology of the Grid
Introduction to SOA and Web Services
The Anatomy and The Physiology of the Grid
Presentation transcript:

Agreement-based Distributed Resource Management Alain Andrieux Karl Czajkowski

OIGS, EdinburghWS-Agreement2 Overview l The Resource Management Problem u Decentralized resource coordination u Resource owner goals vs. application goals l An Open Architecture to Manage Resources u Agreement-based negotiation model u Several scenarios l WS-Agreement (GGF GRAAP-WG) u Status: work in progress u Agreements using OGSI concepts

OIGS, EdinburghWS-Agreement3 Distributed Resource Management 1. Discovery u “What resources are relevant to interest?” u Finds service providers 2. Inspection u “What’s happening to them now?” u Compare/select service providers 3. Agreement u “Will they provide what I need?” u The core Resource Management problem …Process can iterate due to adaptation

OIGS, EdinburghWS-Agreement4 Social/Policy Conflicts l Resource Consumers/Applications Goals u Users: deadlines and availability goals u Applications: need coordinated resources l Localized Resource Owner Goals u Policies distinguish users l Community Goals Emerge As: u Global optimization goals u aggregate user/application and/or resource l Reconcile demands via Agreement

OIGS, EdinburghWS-Agreement5 Early Co-Allocation in Grids l SF-Express (1997-8) u Real-time simulation u 12+ supercomputers, 1400 processors l Required advance reservation u Brokered by telephone! u Practical use requires automation l Complex fault environment u Over 45 minutes to recover from failure u Reservations cannot prevent faults

OIGS, EdinburghWS-Agreement6 Traditional Scheduling l Closed-System Model u Presumption of global owner/authority u Sandboxed applications with no interactions u “Toss job over the fence and wait” l Utilization as Primary Metric u Deep batch queues allow tighter packing u No incentives for matching user schedule l Sub-cultures Counter Site Policies u Users learn tricks for “gaming” their site

OIGS, EdinburghWS-Agreement7 An Open Negotiation Model l Resources in a Global Context u Advertisement and negotiation u Normalized remote client interface u Resource maintains autonomy l Automated Agents Bridge Resources u Drive task submission and provisioning u Coordinate acts across domains l Community-based Mediation u Agents coordinating for collective interest

OIGS, EdinburghWS-Agreement8 Community Schedulers l Individual users u Require service u Have application goals l Community schedulers u Broker service u Aggregate scheduling l Individual resources u Provide service to clients u Have policy autonomy J1J2J3J4J5 R2R3R4R5R6R1 S1S2 J1??J3J4J5J2

OIGS, EdinburghWS-Agreement9 Intermediaries And Policy l Resource virtualization can: u Abstract details of underlying resource(s) u Map between different resource description domains l Policies from different domains influence agreement negotiations with intermediaries Scheduler CommunityClient Application Resource Manager Resource User PolicyResource PolicyCommunity Policy control request respond request respond advertise

OIGS, EdinburghWS-Agreement10 Heterogeneity of Service l Many Kinds of Task u Data: stored file, data read/write u Compute: execution, suspended job l Many Kinds of Resource u Hardware: disks, CPU, memory, networks, display… u Capabilities: space, throughput… l Coordination Problem is much the same

OIGS, EdinburghWS-Agreement11 Specialization: File Transfer l Single goal u Reliable deadline transfer l Specialized scheduler u Brokers basic services u Synthesizes new service l Fault-handling logic l Distributed resources u Storage space u Storage bandwidth u Network bandwidth R1R3R2 J1 S1 J3J2

OIGS, EdinburghWS-Agreement12 Technical Challenges l Complex Security Requirements l Global Scalability u Similar ideals to Internet u Interoperable infrastructure u Policy-configurable for social needs l Permanence or “Evolve in Place” u Cannot take World off-line for service u Over time: upgrade, extend, adapt u Accept heterogeneity

OIGS, EdinburghWS-Agreement13 WS-Agreement Components AgreementFactory Agreement 1 Agreement 2 (binds)

OIGS, EdinburghWS-Agreement14 WS-Agreement Model l Generic/extensible negotiation model u Agreement wraps domain-specific terms u Agreement supports extensible monitoring l Reuse OGSI mechanisms u Specializes ogsi:Factory pattern u Flexible lifetime negotiation for Agreements u ServiceData for monitoring/introspection

OIGS, EdinburghWS-Agreement15 Negotiation Interfaces l AgreementFactory u Persistent service u Ex: façade to scheduler(s) u Creates Agreement services l Agreement u Transient service u Ex: job entry virtualized into a service u Encapsulates state of negotiation l Terms, service status, relationship to other Agreements u Lifetime maps to lifetime of “terms of service”

OIGS, EdinburghWS-Agreement16 Two-level Negotiation l AgreementFactory::createService() u Coarse-grained u Conventional fault/response model u Batch negotiation of complex terms u Idiom: enables one-shot job submission l Agreement::renegotiate() u Fine-grained u Allows complex multi-message negotiation u Admits adaptation of provisioning terms

OIGS, EdinburghWS-Agreement17 Agreement-based Jobs l Agreement represents “queue entry” u Commitment with job parameters etc. l Job structure l Wide range of QoS guarantees u Point for monitoring/control of job l Service is the Job computation u Agreement-specific computation u May or may not communicate with clients l Advance Reservation is “pre-agreement” u Facilitates future job negotiation

OIGS, EdinburghWS-Agreement18 Agreement Terms l Real Agreements mix-in domain terms u Composed by logical grouping u Combined with negotiability mark-up l Each domain term brings a semantics u Unambiguous service-provisioning concept u Y=“amount of RAM allocable to process” l Agreement contextualizes domain term u (Y > 512 MB) AND (Y < 1024 MB)

OIGS, EdinburghWS-Agreement19 The End l WS-Agreement is just beginning u GRAAP-WG at GGF u Work on core negotiation model u Work on reusable term meta-language l Domain Terms needed u Job submission u Data management u Accounting/Economic trading? u …