les robertson cernit 1 - PowerPoint PPT Presentation

Loading...

PPT – les robertson cernit 1 PowerPoint presentation | free to view - id: 145735-MmVjZ



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

les robertson cernit 1

Description:

No recording - 5 days non-stop. 1800 MBytes/s sustained (milestone: 1000 Mbytes/s) ... Recording to CASTOR - 4.5 days non-stop. Data to disk: total data volume : ... – PowerPoint PPT presentation

Number of Views:34
Avg rating:3.0/5.0
Slides: 28
Provided by: lesr150
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: les robertson cernit 1


1
LHC Computing Grid Project
  • Status Report Computing RRB
  • 22 October 2002
  • Les Robertson, CERN
  • les.robertson_at_cern.ch
  • http//www.cern.ch/lcg

2
SC2 Requirements Specification status of RTAGs
  • On applications final report
  • data persistency 05apr02
  • software support process 06may02
  • mathematical libraries 02may02
  • detector geometry description 11oct02
  • Monte Carlo generators 11oct02
  • applications architectural blueprint 11oct02
  • Detector simulation scheduled nov
  • On Fabrics
  • mass storage requirements 03may02
  • On Grid technology and deployment area
  • Grid technology use cases 07jun02
  • Regional Center categorization 07jun02
  • Current status of RTAGs (and available reports)
    on www.cern.ch/lcg/sc2

matthias.kasemann_at_fnal.gov
3
Work Planning Status
  • Launch Workshop at CERN 11-15 March 2002
  • set the scope and priorities for the project
  • High level planning paper prepared and presented
    to LHCC in July
  • see www.cern.ch/lcg/peb ? Status of High Level
    Planning
  • planning evolving rapidly aim to have a formal
    WBS plan by end 2003
  • Formal work plans agreed for
  • Data Persistency
  • Support for the Software Process Infrastructure
  • Mass Storage
  • Work plans in preparation
  • Grid use cases
  • Mathematical Libraries
  • LHC Global Grid Service
  • Scope and scale set by Review of LHC Computing
  • First service definition in preparation

matthias.kasemann_at_fnal.gov
4
Project Execution
  • Four areas
  • Applications
  • Grid Technology
  • Fabric Management
  • Grid Deployment

5
Project Execution Board
  • Decision taking - as close as possible to the
    work - by those who will be responsible for the
    consequences
  • Two bodies set up to coordinate take decisions
  • Architects Forum
  • software architect from each experiment and the
    application area manager
  • makes common design decisions and agreements
    between experiments in the applications area
  • supported by a weekly applications area meeting
    open to all participants
  • Grid Deployment Board
  • representatives from the experiments and from
    each country with an active Regional Centre
    taking part in the LCG Grid Service
  • forges the agreements, takes the decisions,
    defines the standards and policies that are
    needed to set up and manage the LCG Global Grid
    Services
  • coordinates the planning of resources for physics
    and computing data challenges

6
Applications area
  • Three active projects
  • Software Process and Infrastructure (SPI)
  • Persistency Framework (POOL)
  • first general release due mid-2003
  • Math Libraries.
  • Common staffing
  • Applications projects will integrate staff from
    experiments, IT and EP Divisions
  • Already achieved with POOL project
  • EP Division setting up central applications group
  • Migration of key staff to building 32
  • Future activities
  • core framework services
  • simulation
  • detector description
  • event generators
  • analysis tools

7
Applications Support Human Resource Assignments
EP/Experiments includes people from external
institutes - this is expected to increase
as further common activities are agreed New
software support group being set up in EP Division
8
Grid Technology in LCG
  • This area of the project is concerned with
  • ensuring that the LCG requirements are known to
    current and potential Grid projects and industry
  • influencing plans and priorities of supplier
    projects
  • negotiating support for tools developed by Grid
    projects
  • developing a plan to supply solutions that do not
    emerge from other sources
  • BUT this must be done with caution
  • avoid HEP-SPECIAL solutions
  • timely migration to emerging standards

9
Fabrics Area
  • CERN prototype system
  • expanded to 400 systems, 50 TeraBytes of disk
  • mass storage performance being expanded to 350
    MB/sec
  • Prototype used for
  • testbeds for Grid middleware
  • computing data challenges, including ATLAS filter
    farm tests
  • high performance data intensive cluster
  • needed for ALICE data recording challenges
  • will be upgraded with Enterasys 10 Gbit Ethernet
    switch
  • extension of LXBATCH for physics data challenges
  • Next year - revised costing for Phase 2 at CERN
  • Revised trigger and event size data
  • New technology review nearing completion PASTA
    III

10
experience-weighted FTEs
Applications area includes staff from external
institutes Other areas only includes staff at
CERN - major external
contributions from
-- Grid Projects --
Regional Centres - and in
future for Grid infrastructure operations
11
Data Challenges in 2002
12
6 million events 20 sites
13
grid tools used at 11 sites
Alois.Putzer_at_cern.ch
14
ADC IV performances Period 1
  • Event building with flat data traffic
  • No recording - 5 days non-stop
  • 1800 MBytes/s sustained (milestone 1000
    Mbytes/s)
  • Event building and data recording with ALICE-like
    data traffic
  • Recording to CASTOR - 4.5 days non-stop
  • Data to disk
  • total data volume 140 Tbytes
  • 350 MBytes/s sustained (milestone 300
    MBytes/s)

15
RUNNING PHYSICS DATA CHALLENGES WITH ALIEN
AliEn
Production Status
_at_GRID
15100 jobs, 12CPUh/job,
1GB output/job
up to 450 concurrently running jobs
9/30/02
Predrag
.
Buncic
_at_
cern
.
ch
3
16
CERN - Computing Challenges - J. Closier
17
Integrated services from distributed resources
  • This years simulation campaigns have used 10-20
    computer centres as integrated computing
    facilities
  • The short-term target is to introduce Grid
    Technology to reduce manpower costs for
    simulation campaigns
  • BUT
  • the longer-term goal is to learn how to do
    distributed ANALYSIS

18
The LHC Computing Environment

Multi-Tier Model
  • Tier 0 CERN
  • data recording and reconstruction, repository for
    all data.
  • Tier 1 CERN and a small number of centres
  • full range of services including managed mass
    storage, user support and high bandwidth
    networking
  • full copy of the event summary data (ESD), sample
    of the raw data
  • full range of analysis activities, with emphasis
    on data-intensive batch processing
  • Tier 2
  • reliable batch and interactive services,
    supported by good networking to Tier 1 centres
  • substantial data storage - for analysis and
    simulation
  • Tier 3
  • Local facilities, with the emphasis on
    interactive analysis and simulation.

19
Motivation for the Multi-Tier Model
  • integrate CERN, national and regional computing
    resources
  • exploit expertise in regional centres
  • optimise network bandwidth
  • local control over resources

20
  • Data comes from centres reporting resources to
    the Grid Deployment Board
  • This is a subset of the centres providing
    capacity for physics data challenges
  • Some of the capacity will be used for non-LHC
    experiments

21
  • Source
  • Review of LHC Computing (Feb. 2001)
  • CERN numbers scaled down by Task Force 1 (April
    2002)
  • 2005 Tier 01 Regional Centre planning is for
    6,600 KSI2000 ? lt 10 of capacity
    required in 2008

22
Pilot LHC Global Grid Service
target June 2003
LCG-1 opens as a reliable Global Grid Service
  • offering 24 hour, 7 day a week availability,
  • initially about ten Regional Centres in Europe,
    Asia and North America
  • provides a batch service for all four experiments
    for event production and analysis of the
    simulated data
  • the middleware deployed drawn from the European
    US toolkits
  • provides a solid service for some part of data
    challenges operating in 2H03

23
Fully operational LCG-1 Service Distributed
production environment
target November 2003
  • During its first 6 months, LCG-1 evolves in
    terms of
  • reliability
  • availability
  • functionality
  • number of regional centres, capacity
  • Set up operations centre and user support
    infrastructure
  • looking for external institutes to take on these
    responsibilities
  • Capacity and functionality sufficient for the
    Data Challenges scheduled for 2004

24
CERN Computer Centre Upgrade
  • LHC computing requires additional power and air
    conditioning capacity in B513.
  • Following studies in 2000/2001, the following
    plan was developed
  • Convert the tape vault to a machine room area in
    2002
  • Use this space from 2003, both for new equipment
    and to empty part of the existing machine room
  • Upgrade the electrical distribution in the
    existing machine room during 2003-2005, using the
    vault space as a buffer.
  • Create a dedicated substation to meet power
    needs.
  • For air conditioning reasons, the vault should be
    used for bulky equipment with low heat
    dissipation.
  • e.g. Tape robotics.

25
From Tape vault
to computer room
26
Computer Centre Upgrade Next Steps
  • From October 14th
  • Migrate equipment from the machine room to the
    vault.
  • Robots to move from January
  • Subject to Funding
  • Upgrade the Machine Room electrical distribution
    infrastructure from June 2003
  • Start construction of a new substation for the
    computer centre early in 2003.
  • Installation plan re-scheduled now
    extends into Phase 2

27
Summary
  • Human resources are in good shape
  • staff build-up faster than hoped for
  • specification of formal requirements proceeding
    well
  • next 3-4 months ? define the detailed work plan
    for Phase 1
  • optimistic about effective work-sharing between
    CERN, other institutes
  • Target of deploying the pilot LHC Grid in 2003 is
    ambitious
  • not enough experience with Grid middleware?
  • hard to catch up with the scale of the 2002
    physics data challenges
  • large energetic community ? many different
    opinions, goals
  • Materials funding at CERN is a problem
About PowerShow.com