Some ideas for the future

Slides:



Advertisements
Similar presentations
Optimizing Compilers for Modern Architectures Allen and Kennedy, Chapter 13 Compiling Array Assignments.
Advertisements

Computing Lectures Introduction to Ganga 1 Ganga: Introduction Object Orientated Interactive Job Submission System –Written in python –Based on the concept.
CMS HLT production using Grid tools Flavia Donno (INFN Pisa) Claudio Grandi (INFN Bologna) Ivano Lippi (INFN Padova) Francesco Prelz (INFN Milano) Andrea.
Workload Management Workpackage Massimo Sgaravatto INFN Padova.
1 Friday, September 29, 2006 If all you have is a hammer, then everything looks like a nail. -Anonymous.
Order-Independent Texture Synthesis Li-Yi Wei Marc Levoy Gcafe 1/30/2003.
GRID Workload Management System Massimo Sgaravatto INFN Padova.
1 Brief Announcement: Distributed Broadcasting and Mapping Protocols in Directed Anonymous Networks Michael Langberg: Open University of Israel Moshe Schwartz:
Status of Globus activities within INFN (update) Massimo Sgaravatto INFN Padova for the INFN Globus group
 1. Introduction  2. Development Life-Cycle  3. Current Component Technologies  4. Component Quality Assurance  5. Advantages and Disadvantages.
SYSTEM ANALYSIS AND DESIGN
Workflow Management in Condor Gökay Gökçay. DAGMan Meta-Scheduler The Directed Acyclic Graph Manager (DAGMan) is a meta-scheduler for Condor jobs. DAGMan.
Track 1: Cluster and Grid Computing NBCR Summer Institute Session 2.2: Cluster and Grid Computing: Case studies Condor introduction August 9, 2006 Nadya.
Workload Management WP Status and next steps Massimo Sgaravatto INFN Padova.
DataGrid WP1 Massimo Sgaravatto INFN Padova. WP1 (Grid Workload Management) Objective of the first DataGrid workpackage is (according to the project "Technical.
INFSO-RI Enabling Grids for E-sciencE Workload Management System Mike Mineter
INFSO-RI Enabling Grids for E-sciencE DAGs with data placement nodes: the “shish-kebab” jobs Francesco Prelz Enzo Martelli INFN.
Grid Workload Management Massimo Sgaravatto INFN Padova.
Ganga A quick tutorial Asterios Katsifodimos Trainer, University of Cyprus Nicosia, Feb 16, 2009.
EGEE-II INFSO-RI Enabling Grids for E-sciencE EGEE and gLite are registered trademarks Provenance Challenge gLite Job Provenance.
Grid Computing at Yahoo! Sameer Paranjpye Mahadev Konar Yahoo!
Documentation Dr. Andrew Wallace PhD BEng(hons) EurIng
Technology Mapping. 2 Technology mapping is the phase of logic synthesis when gates are selected from a technology library to implement the circuit. Technology.
Virtual Data Management for CMS Simulation Production A GriPhyN Prototype.
WebFlow High-Level Programming Environment and Visual Authoring Toolkit for HPDC (desktop access to remote resources) Tomasz Haupt Northeast Parallel Architectures.
Summary from WP 1 Parallel Section Massimo Sgaravatto INFN Padova.
Grid Compute Resources and Job Management. 2 Grid middleware - “glues” all pieces together Offers services that couple users with remote resources through.
Joshua Adadevoh ECE-3551 Project Audio Signal Encryption.
EGEE-II INFSO-RI Enabling Grids for E-sciencE Practical using WMProxy advanced job submission.
7.1 Operating Systems. 7.2 A computer is a system composed of two major components: hardware and software. Computer hardware is the physical equipment.
Advanced Higher Computing Science The Project. Introduction Worth 60% of the total marks for the course Must include: An appropriate interface using input.
16 th Geant4 Collaboration Meeting SLAC, September 2011 P. Mato, CERN.
Topological Sort. Sorting technique over DAGs (Directed Acyclic Graphs) It creates a linear sequence (ordering) for the nodes such that: –If u has an.
Practical using C++ WMProxy API advanced job submission
Investigating System Requirements
GridOS: Operating System Services for Grid Architectures
Workload Management Workpackage
Duncan MacMichael & Galen Deal CSS 534 – Autumn 2016
GWE Core Grid Wizard Enterprise (
Inferring Temporal Order of Images from 3D Structure
Michael Langberg: Open University of Israel
The Project Management Framework
P-GRADE and GEMLCA.
Compute Area Marco Cecchi Massimo Sgaravatto
Abstract Major Cloud computing companies have started to integrate frameworks for parallel data processing in their product portfolio, making it easy for.
COTS testing Tor Stålhane.
Globus Job Management. Globus Job Management Globus Job Management A: GRAM B: Globus Job Commands C: Laboratory: globusrun.
CPSC-608 Database Systems
DUCKS – Distributed User-mode Chirp-Knowledgeable Server
MSc in Artificial Intelligence Student: Hsiang-Ling Kuo
Topological Sort.
Topological Sort.
Topological Sort Neil Tang 03/02/2010
Topological Ordering Algorithm: Example
Topological Sort CSE 373 Data Structures Lecture 19.
Initial job submission and monitoring efforts with JClarens
CICC Project Meeting Developed two web services (VotableToTabbedFileService – convert votable to tabbed file) and (ExtractionFromSingleVotableService –
Directed Acyclic Graphs && Topological Sorting
The PlayStation Example
Wide Area Workload Management Work Package DATAGRID project
Program Design Language (PDL)
Operating System Introduction.
Topological Ordering Algorithm: Example
a middleware implementation
Topological Ordering Algorithm: Example
GRID Workload Management System for CMS fall production
Introduction to Operating Systems
Operating System Overview
Topological Ordering Algorithm: Example
ReStore: Reusing Results of MapReduce Jobs
Presentation transcript:

Some ideas for the future Francesco.Giacomini@cnaf.infn.it WP1 Meeting Milano, 11th September 2001

Overview Complex jobs Dependency on Condor Use of existing frameworks

Complex Jobs: Definition A complex job is a set of jobs with dependencies among them. Dependencies are temporal and are given, for example, by I/O requirements. subjob B needs as input the output of subjob A The definition is very vague and generic.

Complex Jobs: (initial) Assumptions Dependencies are not cyclical Otherwise subjobs have to execute concurrently (co-allocation problem) The definition is not recursive A subjob cannot be a complex job Otherwise the semantics of a dependency is more complicated

Complex Jobs: Example A B F C D E The complex job can be represented as a Direct Acyclical Graph (DAG)

Complex Jobs: Simple Solution Represent a DAG with a nested ClassAd (?) “Linearise” the DAG for the example: A C D B E F use a “topological sort” algorithm the solution is not unique For each subjob send it to the Resource Broker independently Data representing a dependency is saved in a “temporary” and “anonymous” file

Complex Jobs: Other Issues Job identifiers Interaction with LB What to do in case a subjob fails Various types of dependencies

Dependency on Condor In my opinion we should get rid as soon as possible of our dependency on those parts of Condor which will not likely become Open Source (TM) Starting with Condor-G job submission is not particularly difficult directly on top of Globus job monitoring should be done based on info in the LB

Implementing ComputingElement Implemented the ComputingElement interface directly on top of Globus, so that one can do: ComputingElement ce(“pcfg.cnaf.infn.it”); Job job; job.set_attribute(“executable”, “/bin/sleep”); job.set_attribute(“arguments”, “10”); ce.submit(job); ce.cancel(job);

Using Existing Frameworks We should evaluate and possibly adopt existing (OO) application frameworks, instead or reinventing the wheel there was a comment on this from one of the D1.2 reviewers One possibility is the Adaptive Communication Environment (ACE) (www.cs.wustl.edu/~schmidt/ACE.html)