June 25, 2015 1 GrenchMark: A synthetic workload generator for Grids KOALA Workshop A. Iosup, H. Mohamed, D.H.J. Epema PDS Group, ST/EWI, TU Delft.

Slides:



Advertisements
Similar presentations
European Research Network on Foundations, Software Infrastructures and Applications for large scale distributed, GRID and Peer-to-Peer Technologies Experiences.
Advertisements

7 april SP3.1: High-Performance Distributed Computing The KOALA grid scheduler and the Ibis Java-centric grid middleware Dick Epema Catalin Dumitrescu,
Automating with Open Source Testing Tools Corey McGarrahan rSmart 01-July-08.
GRID INTEROPERABILITY USING GANGA Soonwook Hwang (KISTI) YoonKee Lee and EunSung Kim (Seoul National Uniersity) KISTI-CCIN2P3 FKPPL Workshop December 1,
June 1, Inter-Operating Grids through Delegated MatchMaking Alexandru Iosup, Dick Epema PDS Group, TU Delft, NL Todd Tannenbaum, Matt Farrellee,
June 1, GrenchMark : Towards a Generic Framework for Analyzing, Testing, and Comparing Grids ASCI Conference 2006 A. Iosup, D.H.J. Epema PDS Group,
June 2, GrenchMark : A Framework for Analyzing, Testing, and Comparing Grids CCGrid 2006 A. Iosup, D.H.J. Epema PDS Group, ST/EWI, TU Delft.
June 3, ServMark A Hierarchical Architecture for Testing Grids Santiago, Chile A. Iosup, H. Mohamed, D.H.J. Epema PDS Group, ST/EWI, TU Delft C.
June 3, 2015 Synthetic Grid Workloads with Ibis, K OALA, and GrenchMark CoreGRID Integration Workshop, Pisa A. Iosup, D.H.J. Epema Jason Maassen, Rob van.
2 nd GADA Workshop / OTM 2005 Conferences Eduardo Huedo Rubén S. Montero Ignacio M. Llorente Advanced Computing Laboratory Center for.
Inter-Operating Grids through Delegated MatchMaking Alexandru Iosup, Dick Epema, Hashim Mohamed,Mathieu Jan, Ozan Sonmez 3 rd Grid Initiative Summer School,
Workload Management Workpackage Massimo Sgaravatto INFN Padova.
DAS-3/Grid’5000 meeting: 4th December The KOALA Grid Scheduler over DAS-3 and Grid’5000 Processor and data co-allocation in grids Dick Epema, Alexandru.
1 A Performance Study of Grid Workflow Engines Alexandru Iosup and Dick Epema PDS Group Delft University of Technology The Netherlands Corina Stratan Parallel.
1 Trace-Based Characteristics of Grid Workflows Alexandru Iosup and Dick Epema PDS Group Delft University of Technology The Netherlands Simon Ostermann,
GRID Workload Management System Massimo Sgaravatto INFN Padova.
June 25, GrenchMark: Synthetic workloads for Grids First Demo at TU Delft A. Iosup, D.H.J. Epema PDS Group, ST/EWI, TU Delft.
June 28, Resource and Test Management in Grids Rapid Prototyping in e-Science VL-e Workshop, Amsterdam, NL Dick Epema, Catalin Dumitrescu, Hashim.
June 29, Grenchmark: A workload generator for Grid schedulers First Demo at TU Delft A. Iosup, D.H.J. Epema PDS Group, ST/EWI, TU Delft.
4 december, The Distributed ASCI Supercomputer The third generation Dick Epema (TUD) (with many slides from Henri Bal) Parallel and Distributed.
Workload Management Massimo Sgaravatto INFN Padova.
University of Dortmund June 30, On Grid Performance Evaluation using Synthetic Workloads JSSPP 2006 Alexandru Iosup, Dick Epema PDS Group, ST/EWI,
July 13, GrenchMark: A workload generator for Grids Demo at TU Delft A. Iosup, D.H.J. Epema PDS Group, ST/EWI, TU Delft.
July 13, “How are Real Grids Used?” The Analysis of Four Grid Traces and Its Implications IEEE Grid 2006 Alexandru Iosup, Catalin Dumitrescu, and.
Euro-Par 2008, Las Palmas, 27 August DGSim : Comparing Grid Resource Management Architectures Through Trace-Based Simulation Alexandru Iosup, Ozan.
© Company Confidentialwww.itcinfotech.com Business Case for Test Automation S.Janardhanan Chief Technology Officer ITC Infotech India Limited Business.
DIRAC API DIRAC Project. Overview  DIRAC API  Why APIs are important?  Why advanced users prefer APIs?  How it is done?  What is local mode what.
August 28, Performance Analysis of Cloud Computing Services for Many-Tasks Scientific Computing Berkeley, CA, USA Alexandru Iosup, Nezih Yigitbasi,
Euro-Par 2007, Rennes, 29th August 1 The Characteristics and Performance of Groups of Jobs in Grids Alexandru Iosup, Mathieu Jan *, Ozan Sonmez and Dick.
1 Cloud Computing Research at TU Delft – A. Iosup Alexandru Iosup Parallel and Distributed Systems Group Delft University of Technology The Netherlands.
1 EuroPar 2009 – POGGI: Puzzle-Based Online Games on Grid Infrastructures POGGI: Puzzle-Based Online Games on Grid Infrastructures Alexandru Iosup Parallel.
Workload Management WP Status and next steps Massimo Sgaravatto INFN Padova.
Scheduling of Parallel Jobs In a Heterogeneous Multi-Site Environment By Gerald Sabin from Ohio State Reviewed by Shengchao Yu 02/2005.
Nightly Releases and Testing Alexander Undrus Atlas SW week, May
50mm Telescope ACS Course Garching, 15 th to 19 th January 2007 January 2007Garching.
Service Architecture of Grid Faults Diagnosis Expert System Based on Web Service Wang Mingzan, Zhang ziye Northeastern University, Shenyang, China.
03/27/2003CHEP20031 Remote Operation of a Monte Carlo Production Farm Using Globus Dirk Hufnagel, Teela Pulliam, Thomas Allmendinger, Klaus Honscheid (Ohio.
Through the development of advanced middleware, Grid computing has evolved to a mature technology in which scientists and researchers can leverage to gain.
3-2.1 Topics Grid Computing Meta-schedulers –Condor-G –Gridway Distributed Resource Management Application (DRMAA) © 2010 B. Wilkinson/Clayton Ferner.
Contents 1.Introduction, architecture 2.Live demonstration 3.Extensibility.
Belle MC Production on Grid 2 nd Open Meeting of the SuperKEKB Collaboration Soft/Comp session 17 March, 2009 Hideyuki Nakazawa National Central University.
LCG Middleware Testing in 2005 and Future Plans E.Slabospitskaya, IHEP, Russia CERN-Russia Joint Working Group on LHC Computing March, 6, 2006.
1 Challenge the future KOALA-C: A Task Allocator for Integrated Multicluster and Multicloud Environments Presenter: Lipu Fei Authors: Lipu Fei, Bogdan.
Grid Workload Management Massimo Sgaravatto INFN Padova.
Evaluation of Agent Teamwork High Performance Distributed Computing Middleware. Solomon Lane Agent Teamwork Research Assistant October 2006 – March 2007.
Running a Scientific Experiment on the Grid Vilnius, 13 rd May, 2008 by Tomasz Szepieniec IFJ PAN & CYFRONET.
OMIS Approach to Grid Application Monitoring Bartosz Baliś Marian Bubak Włodzimierz Funika Roland Wismueller.
LHCb planning for DataGRID testbed0 Eric van Herwijnen Thursday, 10 may 2001.
Tool Integration with Data and Computation Grid GWE - “Grid Wizard Enterprise”
July 11-15, 2005Lecture3: Grid Job Management1 Grid Compute Resources and Job Management.
INFSO-RI JRA 1 Testbed Management Technologies Alain Roy (University of Wisconsin-Madison, USA) ETICS 2 Final Review Brussels - 11 May 2010.
August 30, 2002Jerry Gieraltowski Launching ATLAS Jobs to either the US-ATLAS or EDG Grids using GRAPPA Goal: Use GRAPPA to launch a job to one or more.
INFSO-RI Enabling Grids for E-sciencE ARDA Experiment Dashboard Ricardo Rocha (ARDA – CERN) on behalf of the Dashboard Team.
Software Quality Assurance and Testing Fazal Rehman Shamil.
Tool Integration with Data and Computation Grid “Grid Wizard 2”
STAR Scheduling status Gabriele Carcassi 9 September 2002.
Grid Workload Management (WP 1) Massimo Sgaravatto INFN Padova.
WP1 Status and plans Francesco Prelz, Massimo Sgaravatto 4 th EDG Project Conference Paris, March 6 th, 2002.
STAR Scheduler Gabriele Carcassi STAR Collaboration.
Grid Activities in CMS Asad Samar (Caltech) PPDG meeting, Argonne July 13-14, 2000.
Use of Performance Prediction Techniques for Grid Management Junwei Cao University of Warwick April 2002.
2004 Queue Scheduling and Advance Reservations with COSY Junwei Cao Falk Zimmermann C&C Research Laboratories NEC Europe Ltd.
The EPIKH Project (Exchange Programme to advance e-Infrastructure Know-How) gLite Grid Introduction Salma Saber Electronic.
Testing for patch certification
NGS computation services: APIs and Parallel Jobs
Class project by Piyush Ranjan Satapathy & Van Lepham
A Grid Research Toolbox
On Dynamic Resource Availability in Grids
Resource and Test Management in Grids
Presentation transcript:

June 25, GrenchMark: A synthetic workload generator for Grids KOALA Workshop A. Iosup, H. Mohamed, D.H.J. Epema PDS Group, ST/EWI, TU Delft

June 25, About productivity and efficiency… We now have a good scheduler for the DAS (Koala) How do we make use of it every day?

June 25, Productivity, efficiency 1.Run 1 job run 1 command ET (every time)

June 25, Productivity, efficiency 1.Run 1 job run 1 command ET (every time) 2.Run 10 jobs write 10 lines (once), run 1

June 25, Productivity, efficiency 1.Run 1 job run 1 command ET (every time) 2.Run 10 jobs write 10 lines (once), run 1 3.Run 100 jobs, parameterize runs write 25 lines run 1

June 25, Productivity, efficiency 1.Run 1 job run 1 command ET (every time) 2.Run 10 jobs write 10 lines (once), run 1 3.Run 100 jobs, parameterize runs write 25 lines run 1 4.(Re-)Run (the same) 100 jobs, parameterize runs write 500 lines run 1

June 25, Productivity, efficiency 6.Run “100 times each of: 100 jobs, parameterized, 1 to 5 components per job, preference for 2 and 4 components, 2 to 128 processors per job, preference for 2, 4, 16 and 32 processors, arrival rate Poisson or Alpha-Beta or mixed, stop runs after 15 or 60 minutes, …” write 5000 lines run 1

June 25, Productivity, efficiency 6.Run “100 times each of: 100 jobs, parameterized, 1 to 5 components per job, preference for 2 and 4 components, 2 to 128 processors per job, preference for 2, 4, 16 and 32 processors, arrival rate Poisson or Alpha-Beta or mixed, stop runs after 15 or 60 minutes, …” write 5000 lines run 1 7. … and this was per test / project / article

June 25, Problem statement (revisited) Synthetic worklods are the basis of grid schedulers comparison [our work], and of... Functionality testing in the DAS (Early) performance testing for application development Comparing alternative settings for the DAS Many others … Need a standard tool for synthetic workloads generation and submission

June 25, The GrenchMark framework What’s in a name? grid benchmark → help standardizing the testing procedure, but benchmarks are too early… GrenchMark Easy-to-use tools to create synthetic workloads Easy-to-use tools to submit synthetic workloads Flexible, portable, extensible

June 25, The GrenchMark framework

June 25, The GrenchMark process: Describe $ cat wl-desc.in # File-type: text/wl-spec #Times UnitType SiteType SiteInfo ArrivalTimeDistrOtherInfo 50 ssert non-fixed*:*U(0,100) - 10 cpi-gm fixed fs1;…Poisson(60) NProc=1;2;4;… 35 cpi-g2 fixed *:*Poisson(60) ExternalFile= my.data wl-desc.in describes the workload to be generated $ cat my.data NComp=10/50; 4/20; 2/10; 5/10; 3 NProc=1/10; 2/10; 4/10; 8/10; 16/10; 32/10; 3;… … my.data describes the characteristics of a part of the workload

June 25, $./wl-gen.py wl-desc.in The GrenchMark process: Generate Build directory structure Generate Times x Jobs JDFs 2 Sample run: 95 jobs / 411 components 95 files / 32 directories 300KB data

June 25, The GrenchMark process: Submit $./wl-gen.py wl-desc.in $./wl-submit.py out/wl-to-submit.wl 3 3 TODO Sample run: multi-threaded – submission delay +/- 0.01s 95 JDFs

June 25, GrenchMark status Already done in Python Generator + Globus, KOALA generators + RSL printer Submitter (Koala, Globus, …) Results analyzer (crude) Support for applications Unitary, 3 types: sequential, MPI, Ibis (Java) +35 different applications Composite applications: DAG-based Ongoing work Automated results analyzer Better modeling capabilities

June 25, Extending GrenchMark Motto: Extending GrenchMark is easy! 1.More workload generators a function with a predefined name in a Python module auto-loaded 2.Interface with C/C++, Ruby, Perl, Java, … define your own protocol and write the Python end 3.Interface with other systems (SGE, …) a function with a predefined name in a Python module auto-loaded

June 25, GrenchMark for testing K OALA … K OALA Non-fixed, fixed, semi-fixed jobs Test & evaluate Status: pre-release, supposed stable Workloads with different jobs requirements, inter-arrival rates, co-allocated v. single site jobs… Jobs success rate, K OALA ’s overhead and bottlenecks Results 2 major bugs first day, +10 bugs overall (all fixed) K OALA is officially released (full credit to KOALA developers, 10x for testing with GrenchMark) +5,000 jobs successfully run

June 25, … and DAS-2’s functionality … Already done Evaluate for KOALA + Globus + DAS-2 jobs success rate, turnaround time, middleware overhead, types and sources of errors Sample results 5 workloads 500 jobs A.Iosup, J.Maassen, R.V.van Nieuwpoort, D.H.J.Epema, Synthetic Grid Workloads with Ibis, KOALA, and GrenchMark, In Proc. of CoreGRID IW, Nov ,000 jobs successfully run

June 25, … and for analyzing, testing, and comparing grids Grid system analysis Performance testing / analysis What-if analysis Functionality testing in grid environments System functionality testing Periodic testing Comparing grid settings Single-site vs. co-allocated jobs Unitary vs. composite jobs A.Iosup, D.H.J.Epema, GrenchMark: A Framework for Analyzing, Testing, and Comparing Grids, In CCGrid 2006 (accepted). +2,500 jobs successfully run

June 25, Conclusions GrenchMark generates and submits synthetic workloads easy-to-use, flexible, portable, extensible, … Experience used GrenchMark to test KOALA’s functionality and performance. used GrenchMark to test some DAS/Grid functionality. +5,000 jobs generated and run … and counting. … advertisement Have specific Grid applications types you would like to test? Test with GrenchMark!

June 25, Thank you! Questions? Remarks? Observations? All welcome! Grenchmark [ “grenchmark”] Alexandru IOSUP TU Delft [ “iosup”] Also: H. Mohamed and W. Lammers (K OALA ), K. Verstoep and P. Anita (DAS admins), R. van Nieuwpoort and J. Maassen (Ibis)

June 25, Outline Introduction Grenchmark Using Grenchmark Customizing Grenchmark Extending Grenchmark Conclusions [done] [here]