Open Science Grid.. An introduction

1
Open Science Grid.. An introduction

• Ruth Pordes
• Fermilab

2
Why am I at Maryland today ?

• The US LHC Experiments, which will take data at
  CERN starting next year have a global data
  processing and distribution model based on Grids.
• US CMS (and US ATLAS) are relying on and
  contributing to the Open Science Grid as their
  distributed facility in the US. Maryland is a US
  CMS facility.
• Maryland physics also includes D0 - already
  using Grids for data distribution and processing
  Icecube - already collaborating with Condor in
  Wisconsin. .
• I am attending the Open Grid Forum standards
  meeting in Washington - and yes, I met David
  McNabb

3
LHC Global Data Grid (2007)

• 5000 physicists, 60 countries
• 10s of Petabytes/yr by 2008
• 1000 Petabytes in lt 10 yrs?

CMS Experiment
Online System
CERN Computer Center
0.1 - 1.5 GB/s
Tier 0
10-40 Gb/s
Tier 1
gt10 Gb/s
Tier 2
2.5-10 Gb/s
Tier 3
Tier 4
Physics caches
PCs
4
Computing Facility Tiers in CMS

• Tier-1 One Center in each Geographical region
  of the world. Well provisioned with Petabytes of
  Archival Storage to accept data from and provide
  data to Tier-2s/Tier-Ns.
• Tier-2s Several (in the US 7) University
  facilities that accept responsibility to deliver
  to the CMS experiment Monte Carlo simulations and
  analysis and provide 200 TB of online data
  caches for analysis and mentor the Tier-3s.

• Tier-3s University facilities providing
  processing and online storage for the local
  physics and research groups, and provisioned to
  get data from and send data to Tier-2s.

5
What is Open Science Grid?

• It is a supported US distributed facility
• The OSG Consortium builds and operates the OSG.
• The farms and storage are not owned by OSG - we
  integrate existing resources.
• 5 years of funding 6M/year (35 FTEs) starting
  9/2006 from DOE SciDAC-II and NSF MPS and OCI.
• Cooperation with other distributed facilities /
  Grids to give our research groups transparent
  access to resources they are offered worldwide.

6
The OSG Map Aug-2006
10 SEs 50 CEs (about 30 very active)
Some OSG sites are also on TeraGrid or EGEE.
7
Open Science Grid in a nutshell

• Set of Collaborating Computing Farms - Compute
  Elements.

Commodity Linux Farms Disk. optional MSS
OSG the Wide Area Network
OSG CE gateway
From 20 CPUs in Department Computers To 10,000
CPU SuperComputer
Any University Local Batch System
8

• Set of Collaborating Storage Sites - Storage
  Elements.

OSG the Wide Area Network
Mass Storage Systems And Disk Caches
OSG SE gateway
From 20 GBytes Disk Cache To 4 Petabyte Robotic
Tape Systems
Any Shared Storage
9
OSG Services

• Grid wide monitoring, information, accounting
  services in support of a System.
• Based on
• X509 GSI Security
• Globus Toolkit
• Condor products
• Storage Resource Management (SRM) storage
  interface
• VOMS/Prima/GUMS Role Based Attribute and Access
  Control
• S/w to make it interoperable with Europe and
  TeraGrid.

10
Supported Software Stacks

• Integrated Supported Reference Software Services
• Most Services on Linux PC Gateways -- minimal
  impact on compute nodes.
• Loose coupling between services, heterogeneity in
  releases and functionality.

Independent Collections for Client,
Server, Administrator
11
Middleware Service Principles Me -- My friends
-- The grid

• Middleware

12
Grid - of - Grids

• Inter-Operating and Co-Operating Grids Campus,
  Regional, Community, National, International
• Open Consortium of Virtual Organizations doing
  Research Education

13
Data Transfer for LHC crucial

• Core Goal to Deliver to US LHC LIGO scale in
  next 2 years

1 GigaByte/sec
14
Who is OSG ?

• Large global physics collaborations US ATLAS,
  US CMS, LIGO, CDF, D0, STAR
• Education projects e.g. Mariachi,I2U2.
• Grid technology groups Condor, Globus, SRM, NMI.
• Many DOE Labs and DOE/NSF sponsored University IT
  facilities.
• Partnerships e.g. TeraGrid, European Grids,
  Regional/Campus Grids e.g. Texas, Wisconsin

15
OSG Consortium
16
Current OSG deployment -across integration and
production grids

• 96 Resources

27 Virtual Organizations
17
Smaller VOs
18
Large VOs ...
CMS
CDF
ATLAS
19
Running a Production Grid
20
Running a Production Grid
21
OSG Core Program of Work

• Integration software and systems.
• Operations common support and procedures.
• Inter-Operation across administrative and
  technical boundaries.

22
Release Process (Subway Map)
Gather requirements

Time
Build software
Test
Validation test bed
VDT Release
ITB Release Candidate
Integration test bed
OSG Release
23
What is the VDT?

• A collection of software
• Grid software (Condor, Globus and lots more)
• Virtual Data System (Origin of the name VDT)
• Utilities
• Built for gt10 flavors/versions of Linux
• An easy installation
• Goal Push a button, everything just works
• Two methods
• Pacman installs and configures it all
• RPM installs some of the software, no
  configuration
• A support infrastructure

Build Software
24
What software is in the VDT?

• Security
• VOMS (VO membership)
• GUMS (local authorization)
• mkgridmap (local authorization)
• MyProxy (proxy management)
• GSI SSH
• CA CRL updater
• Monitoring
• MonaLISA
• gLite CEMon
• Accounting
• OSG Gratia

• Job Management
• Condor (including Condor-G Condor-C)
• Globus GRAM
• Data Management
• GridFTP (data transfer)
• RLS (replication location)
• DRM (storage management)
• Globus RFT
• Information Services
• Globus MDS
• GLUE schema providers

25
What software is in the VDT?

• Client tools
• Virtual Data System
• SRM clients (V1 and V2)
• UberFTP (GridFTP client)
• Developer Tools
• PyGlobus
• PyGridWare
• Testing
• NMI Build Test
• VDT Tests

• Support
• Apache
• Tomcat
• MySQL (with MyODBC)
• Non-standard Perl modules
• Wget
• Squid
• Logrotate
• Configuration Scripts
• And More!

26

• VO Registers with with Operations Center.
• User registers with VO.
• Sites Register with the Operations Center.
• VOs and Sites provide Support Center Contact and
  join Operations groups

Were all fun people!
27
The OSG VO

• A VO for individual researchers and users.
• Managed by the OSG itself.
• Where one can learn how to use the Grid!

28
Due diligence to Security

• Risk assessment, planning, Service auditing and
  checking
• Incident response, Awareness and Training,
  Configuration management,
• User access Authentication and Revocation,
  Auditing and analysis. End to end trust in
  quality of code executed on remote CPU
  -signatures?
• Identity and Authorization Extended X509
  Certificates
• OSG is a founding member of the US TAGPMA.
• DOEGrids provides script utilities for bulk
  requests of Host certs, CRL checking etc.
• VOMS extended attributes and infrastructure for
  Role Based Access Controls.

29
Role Based Access Control
30
Training - e.g. Grid Summer Workshop Year 4
Hands on.Technical trainers.Nice Setting
(Padre Island).Students got their own
applications to run on OSG!
31
Network Connectivity

• Use commodity networks - ESNet, Campus LANs
• Well network provisioned sites e.g. connected to
  Starlight to low bandwidth connections e.g.
  Taiwan
• Connectivity ranges from full-duplex, outgoing
  only, to fully behind firewalls.

32
Bridging Campus Grid Jobs - GLOW

• Dispatch jobs from local security, job, storage
  infrastructure and uploading to wide-area
  infrastructure.

33
FermiGrid?

• Interfacing All Fermilab Resources to common
  Campus Infrastructure
• Gateway to Open Science Grid
• Unified and reliable common interface and
  services through one FermiGrid gateway -
  security, job scheduling, user management, and
  storage.
• Sharing Resources
• Policies and Agreements enable fast response to
  changes in resource needs by Fermilab users.
• More information is available at
  http//fermigrid.fnal.gov

34
Access to FermiGrid
FermiGrid Gateway
35
Web sites

• www.opensciencegrid.org
• http//vdt.cs.wisc.edu
• http//osg.ivdgl.org/twiki/bin/view