Research Unit UTIC Laboratory LIPN , UMR 7030 ESSTT, University of Tunis CNRS, University of Paris 13 Outline Outline Introduction (the concept of Desktop Grids) Objectives of the talk How to decentralize Desktop Grid middlewares, p , Two visions in our research group lessons learned and future works BonjourGrid PastryGrid Heithem Abbes, Christophe Cérin, Mohamed Jemni Other visions for the future of Destop Grids p christophe.cerin@lipn.univ-paris13.fr DAAD Summer School on Current Trends in Distributed Systems 26/09/2009 26/09/2009 CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 2 Introduction 1/3 Introduction 1/3 Introduction 2/3 Introduction 2/3 Grid computing : obtaining an infrastructure offering P2P systems have allowed large improvements in the field of file P2P systems have allowed large improvements in the field of file computing power for users applications. sharing over Internet. Coordination between machines during the execution of an Gnutella, Kazaa and Freenet application. Centralized or hierarchical architectures (Globus, Glite, Condor). � No scalability Decentralized architecture � Complicated procedure of installation No coordination between machines � Complicated configuration phase for an ordinary user Complicated configuration phase for an ordinary user Examples : TeraGrid, EGEE and OpenScience Grid (OSG) : CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 3 CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 4
The history of Desktop Grids by Franck Cappello y p y pp Introduction 3/3 Introduction 3/3 Desktop Grid led the community to build computing systems b based on voluntary machines. d l hi Current systems use Master/Worker model United Devices, BOINC, PLANETLAB, XtremWeb, Condor , , , , � A � Applications domain li ti d i ◦ Global climate prediction (BOINC) ◦ Search for extraterrestrial intelligence (SETI@Home) ◦ Cosmic rays study (XtremWeb). Cosmic rays study (XtremWeb) � Demonstrate the potential of Desktop Grid � To suffer from being hardly scalable due to centralized control � To suffer from being hardly scalable due to centralized control Date � To rely on permanent administrative staff who guarantees the master operation CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 5 CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 6 The history of Desktop Grids by Franck Cappello y p y pp The history of Desktop Grids by Franck Cappello y p y pp CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 7 CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 8
Objectives of this talk Objectives of this talk BonjourGrid Middleware BonjourGrid Middleware To offer a collaborative decentralized and multi-coordinators To offer a collaborative, decentralized and multi coordinators platform To offer a comprehensive survey of (some) hot topics in Desktop Grids To build an infrastructure which does not depend on a central element. l t No static coordinator To illustrate with innovative middleware To create coordinators in a dynamic, automatic way and To create coordinators in a dynamic automatic way and without system administrator intervention To explain views from other researchers Each coordinator asks and seeks, in a decentralized way, idle y machines to participate to the execution of a given application To motivate people to join our community Pluggable computing systems CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 9 CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 10 BonjourGrid : Basic design (1/3) BonjourGrid : Basic design (1/3) BonjourGrid : Basic design (2/3) BonjourGrid : Basic design (2/3) Coordinator Coordinator Workers Computing Element (CE) = 1 Coordinator + N Workers CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 11 CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 12
Advantages in using pub/sub systems BonjourGrid : Basic design (3/3) BonjourGrid : Basic design (3/3) A computing element for each user No static coordinator Only 3 primitives in the toolbox: publish, subscribe, browse , User B User D Notions of global state and global/multicast Notions of global state and global/multicast operations Easy to develop applications User C User C User A Coordinator Worker Idle CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 13 CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 14 BonjourGrid’s vision BonjourGrid s vision Fundamental Parts Fundamental Parts A user requests for a computation A user requests for a computation BonjourGrid is composed of three fundamental parts: BonjourGrid is composed of three fundamental parts: He provides tasks graph and codes implementing his distributed A resources discovery protocol application Fully decentralized protocol Fully decentralized protocol He deploys locally a coordinator node and requests for A “computing elements” constructor participants Executes and handles the various tasks of an application pp The coordinator node selects a subset of idle nodes (CPU, The coordinator node selects a subset of idle nodes (CPU (XtremWeb, Condor, Boinc, MPICH) RAM, Cost) A global coordination protocol When the coordinator node finishes application tasks it pp M Manages and controls all resources, services and computing d t l ll i d ti becomes free and returns to the idle status elements � Its worker nodes become idle Does not depend on any specific machine or centralized element CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 15 CTDS 2009 How to decentralize Desktop Grid middlewares, lessons learned and future works 16
Recommend
More recommend
