Open Science Grid one grid among many

1
Open Science Gridone grid among many
Ruth Pordes Fermilab May 3rd 2006
2
of course a special grid
its the people(some of them at the consortium
meeting in Jan 06)
3
With special partners
4
Grid providers serve multiple communities Grid
consumers use multiple grids.

• The Open Science Grid Consortium brings
• the grid service providers - middleware
  developers, cluster, network and storage
  administrators, local-grid communities
• the grid consumers - from global collaborations
  to the single researcher, through campus
  communities to under-served science domains
• into a cooperative to share and sustain a common
  heterogeneous distributed facility in the US and
  beyond.

5

• The Open Science Grid Consortium brings

I am the Executive Director Miron Livny is
Manager of the OSG Distributed Facility Head of
the Condor Project and Virtual Data Toolkit,
Coordinator of US federation in EGEE, member of
EGEE/gLITE design team. Bill Kramer is the Chair
of the Science Council Head of Berkeley Lab
NERSC supercomputing facility. Ian Foster is
co-PI of the OSG Proposal responsible for
Globus and Computer Science research
contributions and partnerships. Harvey Newman
represents Advanced Network project contributions
and collaborations. Alan Blateky is Engagement
Coordinator for new communities.. Experiment
software leadership US ATLAS and US CMS Sw/C
leaders, LIGO, CDF, D0, STAR etc.
6
The OSG Eco-System Bio Interdependence

• With
• international and national infrastructures -
  EGEE, TeraGrid
• a growing number of campus grids - GLOW, GROW,
  GRASE, FermiGrid, Crimson Grid, TIGRE
• the end-user integrated distributed systems -
  LIGO Data Grid, CMS and ATLAS distributed
  analysis systems, Tevatron SAMGrid, and STAR
  Data Grid.

7
What is Open Science Grid?

• High Throughput Distributed Facility
• Shared opportunistic access to existing clusters,
  storage and networks.
• Owner controlled resources and usage policies.
• Supporting Science
• 5 year Proposal submitted to NSF and DOE - should
  hear in June.
• Open and Heterogeneous
• Research groups transitioning from extending
  (legacy) systems to Grids
• Experiments developing new systems.
• Application Computer Scientists looking for Real
  life use of technology, integration, operation.
• University Researchers...

8
What is Open Science Grid?
Blueprint Principles (june 2004)
Preserve Site autonomy and shared Grid use with
local access. VO based Environment and
Services. Recursive principles throughout -
support grid of grids
9
First foremost - delivery to the WLCG schedule
for LHC science
And soon a third Naregi
10
OSG More than a US Grid
Korea
Brazil - (D0, STAR, LHC)
Taiwan - (CDF, LHC)
11
OSG 1 day last week
Routed from Local UWisconsin Campus Grid
LHC
Bioinformatics

• 50 Clusters used locally as well as through the
  grid
• 5 Large disk or tape stores
• 23 VOs
• gt2000 jobs running through Grid

2000 running jobs
Run II
500 waiting jobs
12
The Trend?
OSG 0.4.0 deployment
13
While LHC Physics drives the schedule and
performance envelope
1 GigaByte/sec
14
OSG also Serves other stakeholders

• Gravitational Wave and other legacy Physics exps.
• E.g. From OSG Proposal LIGO With an annual
  science run of data collected at roughly a
  terabyte of raw data per day, this will be
  critical to the goal of transparently carrying
  out LIGO data analysis on the opportunistic
  cycles available on other VOs hardware
• Opportunity to share use of standing army of
  resources
• E.g. Genome Analysis and Database Update system,
• Interfacing existing computing and storage
  facilities and Campus Grids to a common
  infrastructure.
• E.g. FermiGrid Strategy To allow opportunistic
  use of otherwise dedicated resources. To save
  effort by implementing shared services. To work
  coherently to move all of our applications and
  services to run on the Grid.

15
OSG also Serves other stakeholders

• Gravitational Wave and other legacy Physics exps.
• E.g. From OSG Proposal LIGO With an annual
  science run of data collected at roughly a
  terabyte of raw data per day, this will be
  critical to the goal of transparently carrying
  out LIGO data analysis on the opportunistic
  cycles available on other VOs hardware
• Opportunity to share use of standing army of
  resources
• E.g. From OSG news Genome Analysis and Database
  Update system,
• Interface existing computing and storage
  facilities and Campus Grids to a common
  infrastructure.
• E.g. FermiGrid Strategy To allow opportunistic
  use of otherwise dedicated resources. To save
  effort by implementing shared services. To work
  coherently to move all of our applications and
  services to run on the Grid.

3 Examples of Interoperation
16
Grid Laboratory of Wisconsin (GLOW)
17
GLOW to OSG and the Football Pool problem

• Routing jobs from lan-grid local security, job,
  storage infrastructure and to wan-grid.
• Middleware development from CMS DISUN outreach
  program.
• The goal of the application is to determine the
  smallest "covering code" of ternary words of
  length six. (Or in the football pool, to
  determine how many lottery tickets one would have
  to buy to guarantee that no more than one
  prediction is incorrect.) Even after decades of
  study, only fairly weak bounds are known on this
  value. Solutions to this problem have
  applications in data compression, coding theory
  and statistical designs.

18
Opportunistic Routing from GLOW to OSG
19
TeraGrid

• Through high-performance network connections,
  TeraGrid integrates high-performance computers,
  data resources and tools, and high-end
  experimental facilities around the (US) country.
• CDF MonteCarlo jobs running on Purdue TeraGrid
  resource able to access OSG data areas and be
  accounted to both Grids.

http//www.nsf.gov/news/news_images.jsp?cntn_id10
4248orgOLPA
20
Genome Analysis and Database Update system

• Runs across TeraGrid and OSG. Uses the Virtual
  Data System (VDS) workflow provenance.
• Pass through public DNA and protein databases for
  new and newly updated genomes of different
  organisms and runs BLAST, Blocks, Chisel. 1200
  users of resulting DB.
• Request 1000 CPUs for 1-2 weeks. Once a month,
  every month. On OSG at the moment gt600CPUs and
  17,000 jobs a week.

21
Interoperation Commonality with EGEE

• OSG sites publish Information to WLCG BDII so
  Resource Brokers can route jobs.
• Operations
• Security
• Middleware

22
OSG Middleware Layers

VO SpecificServices Interfaces
LHCServices Interfaces
Tevatron CDF, D0Interfaces
LIGOData Grid
Applications
OSG Release Cache VDT Configuration,
Validation, VO management,
Virtual Data Toolkit (VDT) Common Services NMI
VOMS, MonaLisa, Clarens, AuthZ etc
Infrastructure
NSF Middleware Initiative (NMI) Condor, Globus,
Myproxy
23
OSG Middleware Layers

VO SpecificServices Interfaces
LHCServices Interfaces
Tevatron CDF, D0Interfaces
LIGOData Grid
Applications
OSG Release Cache VDT Configuration,
Validation, VO management,
Virtual Data Toolkit (VDT) Common Services NMI
VOMS, MonaLisa, Clarens, AuthZ etc
Infrastructure
NSF Middleware Initiative (NMI) Condor, Globus,
Myproxy
24
Virtual Data Toolkit V1.3.10b - a collection of
components to integrate to a Distributed System
Easy to download, install and use.
Common with EGEE/WLCG
25
Virtual Data Toolkit V1.3.10b - a collection of
components to integrate to a Distributed System
Easy to download, install and use.
EGEE/LCG at VDT 1.2.4? OSG prepared to help
facilitate upgrade if needed.
26
OSG Program of Work

• Sustained, Robust Distributed Facility
• Operations Integration
• Security
• Software Releases
• Engagement
• Education, Training and Outreach
• Science Driven Extensions
• No developments in OSG - so dependent on external
  projects for extended and new middleware.
• Driven by schedule of stakeholders.
• Will be actively monitoring/inputting to Globus
  CDIGS roadmap and campaigns. Participate in Grid
  Interoperability Now (GIN) when effort available.
  
• Collaborate with gLITE whereever possible.

27
The Vision
S U R F
the Grid
28
Secure
Usable
Reliable
Fast
29
Secure

• Apply the NIST process
• Management - Risk assessment, planning, Service
  auditing and checking,
• Operational - Incident response, Awareness and
  Training, Configuration management,
• Technical - Authentication and Revocation,
  Auditing and analysis. End to end trust in
  quality of code executed on remote CPU
  -signatures?
• Controls.
• http//csrc.nist.gov/index.html

30
Usable me, my friends, the grid (Frank
Wüerthwein)
(3) Rich set of Virtual Organization Services.
Virtual Organization Services, Systems, Admins
31
New Services coming in OSG

• Pull Mode Pilot Jobs just in time binding of
  job to site (Panda, GlideCAF, Condor-C) VO
  downloaded executables subject to site
  authorization and security callouts/services. Use
  of gLITE GLEXEC.
• Virtual Machine based Workspaces VO/Globus
  workspaces encapsulate services.
• Worker Nodes need not have access to the WAN
  use of Condor Grid Connection Broker (GCB)
• Resource Selection based on ClassAds gLITE
  CEMON.
• Move to WS GT4 Tests of WS Gram with CMS CRAB
  jobs sent Globus back to development table.Next
  MDS4.
• Incremental upgrades where sensible. For
  HeadNodes (edge services) cleaner, we may make
  it a requirement, to replicate service and
  support both in parallel.
• Accounting Condor meter possibility to share
  probes/meters with gLITE. Agreement on GGF Usage
  Record - needs extending. Joint EGEE, OSG,
  TeraGrid monthly phone-calls.

32
Digression..Accounting What is an OSG
Job?Resources can be on Multiple Grids
MyApplication
Job SubmissionCondor-G
EGEE
OSG
Job Counted on OSG EGEE
33
On the Campus Grid, On the VO grid, submitted to
the local cluster by resource selector, do work
across multiple grids, consume differing value
MyApplication
Job SubmissionCondor-G
US CMS DISUN Accounting
EGEE
OSG
Job Counted on Campus Grid as well
34
Reliable Central Operations Activities
Grid Exerciser

• Automated validation of basic services and site
  configuration.
• Robots of various kinds.

35
Fast

• Integrating network management into s/w stack
  (LambaStation).
• VO specific resource selection/brokering.
• Include support for persistent VO s/w on sites
  Posix(like) I/O to data at Worker Nodes.
• Tune/Configure/Replicate Headnodes.
• Trying to stay ahead of the needed amount of
  resources - while fully supporting opportunistic
  use. Policy, Priorities, Monitoring.

36
OSG Where to find information

• OSG Web site www.opensciencegrid.org
• Work in progress http//osg.ivdgl.org/twiki/bin/v
  iew/Integration/OverviewGuide
• Virtual Data Toolkit http//vdt.cs.wisc.edu//i
  ndex.html
• News about Grids in Science in Science Grid This
  Week www.interactions.org/sgtw
• OSG Consortium meeting Seattle Aug 21st.

Thank you!