lect20-condor-boinc

lect20-condor-boinc - CMSC 714 Lecture 20 Finding Idle...

Info iconThis preview shows pages 1–13. Sign up to view the full content.

View Full Document Right Arrow Icon
CMSC 714 Lecture 20 Finding Idle Cycles Alan Sussman
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
2 Notes Exam Thursday – on all papers through today – questions? Group project interim report due Monday, Nov. 21
Background image of page 2
Condor and BOINC Distributed and Volunteer Computing Presented by Adam Bazinet
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Condor Developed at the University of Wisconsin-Madison Condor is aimed at High Throughput Computing (HTC) on collections of distributively owned resources Mainly used to scavenge idle CPU cycles from workstations
Background image of page 4
Typical Condor Pool Central Manager master collector schedd startd = ClassAd Communication Pathway = Process Spawned Submit-Only master schedd Execute-Only master startd Regular Node schedd startd master Regular Node schedd startd master Execute-Only master startd
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Condor Daemons condor_master - keeps other daemons running condor_startd - advertises a given resource condor_starter - spawns a remote Condor job condor_schedd - local job scheduler condor_shadow - coordinates with submitted job condor_collector - keeps status of Condor pool condor_negotiator - does all matchmaking
Background image of page 6
Condor Universes Universes are runtime environments for jobs Standard universe Provides checkpointing and remote system calls Application must be re-linked with condor_compile Vanilla universe Instead of with remote system calls, Fles are accessed with N±S/A±S or explicitly transferred to the executing host Other universes: PVM, MPI, Globus, Java, Scheduler
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Matchmaking Matchmaking is Condor’s scheduling mechanism Jobs specify their requirements as a list of attributes and values Resources advertise their capabilities as a list of attributes and values (ClassAds) The condor_negotiator matches jobs to resources using these criteria
Background image of page 8
Condor - A Hunter of Idle Workstations Michael J. Litzkow, Miron Livny, Matt W. Mutka
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Previous Work In three key areas: The analysis of workstation usage patterns The design of remote capacity allocation algorithms The development of remote execution facilities
Background image of page 10
Design Goals Condor is designed to serve users executing long running background jobs on idle workstations Job placement should be transparent Job migration should be supported Fair access to cycles is expected The system should be low overhead
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
The Scheduling Spectrum
Background image of page 12
Image of page 13
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

Page1 / 41

lect20-condor-boinc - CMSC 714 Lecture 20 Finding Idle...

This preview shows document pages 1 - 13. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online