Title: CMS%20Software%20and%20Computing%20FNAL%20Internal%20Review%20of%20USCMS%20Software%20and%20Computing
1CMS Software and ComputingFNAL Internal Review
of USCMS Software and Computing
- David Stickland
- Princeton University
- CMS Software and Computing Deputy Project Manager
2CMS Software and Computing - Overview
- Outline
- Status of Software / Computing Milestones
- CMS Software Architecture development plans
- CMS Computing plans
- CMS Reorganization of Software and Computing
Project - Key issues in 2001
3Strategic Software Choices
- Modular Architecture (flexible and safe)
- Object-Oriented Framework
- Strongly-typed interface
- Uniform and coherent software solutions
- One main programming language
- One main persistent object manager
- One main operating system
- Adopt standards
- Unix, C, ODMG, OpenGL...
- Use widely spread, well supported products (with
healthy future) - Linux, C, Objectivity, Qt...
- Mitigate risks
- Proper planning with milestones
- Track technology evolution investigate and
prototype alternatives - Verify and validate migration paths have a
fall-back solution ready
4CARFCMS Analysis Reconstruction Framework
Physics modules
Specific Framework
Reconstruction Algorithms
Data Monitoring
Event Filter
Physics Analysis
Generic Application Framework
Calibration Objects
Event Objects
Configuration Objects
CMS adapters and extensions
Utility Toolkit
ODBMS
C standard library Extension toolkit
Geant3/4
CLHEP
Paw Replacement
5Milestones Software
- CMS software development strategy
- First, transition to C, then functionality,
then performance
6Software Life Cycle
Design
Prototypes
Detector
Mass-production
Installation Commissioning
Maintenance Operation
Computing Hardware
Functional Prototype
Fully Functional
Production System
Software Integration
Design
Prototype
Deploy
Cyclic Releases
Test Integrate
2002
2001
2005
2000
2025
7Software Development Phases
- 1 Proof of Concept End of 1998
- Basic functionality
- Very loosely integrated
- 5 Production System
- Online / Trigger Systems 75 ? 100Hz
- Offline Systems few 1015 Bytes / year
- 109 events / year to look for a handful of
(correct!) Higgs - Highly distributed collaboration and resources
- Long lifetime
- 2 Functional Prototype
- More complex functionality
- Integrated into projects
- Preparation for Trigger and DAQ TDRs
- Reality Check 1 Data Challenge
- 3 Fully Functional System
- Complete Functionality
- Integration across projects
- Reality Check 5 Data Challenge
- SW/computing TDR
- Preparation for Physics TDR
Logarithmic
- 4 Pre-Production System
- Reality Check 20 Data Challenge
2002
2001
2004
2003
2005
2000
2025
8Significant Requirements from CMS TDRs or
its not just an evolution to 2005 software
TDR
Software Computing Dec 2002
Trigger Dec 2000
DAQ Dec 2001
Physics Dec 2003
Major Core Software Milestones
2002
2001
2004
2003
2005
2000
2025
9Milestones Data
- Raw data 1 MB / event, 100 Hz ? 1 PB /
year - reconstructed data, physics objects,
calibration data, simulated data, ...
Now moving towards distributed production and
analysis
10Milestones Hardware
TDR and MoU
11Why Worldwide Computing?Regional Center Concept
Advantages
- Common access for Physicists everywhere
- Maximize total funding resources while meeting
the total computing need - Proximity of datasets to appropriate resource
- Tier-n Model
- Efficient use of network bandwidth
- Local gt regional gt national gt international
- Utilizing all intellectual resources
- CERN, national labs, universities, remote sites
- Scientists, students
- Greater flexibility to pursue different physics
interests, priorities, and resource allocation
strategies by region - Systems complexity
- ? Partitioning of facility tasks, to manage and
focus resources
12Milestone Review
- Regional Centres
- Identify initial candidates 06/2000
- Turn on functional centres 12/2002
- Fully operational centres 06/2005
- Central (CERN) systems
- Functional prototype 12/2002
- Turn on initial systems 12/2003
- Fully operational systems 06/2005
- Need to define intermediate working milestones
13CMS Regional Centre Prototypes 2003
- Candidates
- Tier1 Tier2
- Finland - Helsinki
- France CCIN2P3/Lyon ?
- India - Mumbai
- Italy INFN INFN, at least 3 sites
- Pakistan - Islamabad
- Russia Moscow Dubna
- UK RAL ?
- US FNAL Caltech, U.C.-San Diego,Florida,
Iowa, Maryland Minnesota, Wisconsin and
others
14The Grid Services Concept
- Standard services that
- Provide uniform, high-level access to a wide
range of resources (including networks) - Address interdomain issues security, policy
- Permit application-level management and
monitoring of end-to-end performance - Perform resource discovery
- Manage authorization and prioritization
- Broadly deployed (like Internet Protocols)
15Why CMS (HEP) in the GRID?
- GRID Middleware provides a route towards
effective use of distributed resources and
complexity management - GRID design matches the MONARC hierarchical Model
- We (HEP) have some Grid-like tools (hand-made).
The scale of CMS Computing requires a more
professional approach to live for decades. - CMS already participates in relevant GRID
initiatives, e.g. - The Particle Physics Data Grid (PPDG) US
Distributed Data Services and Data Grid System
Prototypes - Grid Physics Network (GriPhyN ) US
Production-Scale Data Grids - DATAGRID EU Middleware development and Real
Applications Test
16Grid Data Management Prototype (GDMP)
- Distributed Job Execution and Data
HandlingGoals - Transparency
- Performance
- Security
- Fault Tolerance
- Automation
Site A
Site B
Submit job
Replicate data
Job writes data locally
Replicate data
- Jobs are executed locally or remotely
- Data is always written locally
- Data is replicated to remote sites
Site C
17Computing Progress
- CMS is progressing towards a coherent distributed
system, to support production and analysis - We need to study the problems and prototyping
the solutions for distributed analysis by
hundreds of users in many countries - Production, via prototypes, will lead to
decisions about the architecture on the basis of
measured performances and possibilities
18CMS Software and Computing
- Evolve the organization to build a complete and
consistent Physics Software - Recognize cross-project nature of key
deliverables - Core Software and Computing CSWC
- More or less what US calls SWC Project
- Physics Reconstruction Selection PRS
- Consolidate Physics Software work between the
detector groups targeted at CMS deliverables (HLT
design, test-beams, calibrations, Physics TDR.. - Trigger and Data Acquisition TRIDAS
- Online Event Filter Farm
19Cross-Project Working Groups
Core Software and Computing
Physics Reconstruction Selection
TRIDAS
Reconstruction Project
Simulation Project
Calibration etc..
Joint Technical Board
20Key issues for 2001
- Choice of baseline for database
- Ramp-up of Grid RD and use in production
activities - Perform HLT data challenge
- ( 500CPUs at CERN 500 offsite, 20 TB)
- Continue work on test beams, and validate
simulation - Validate OSCAR/Geant4 for detector performance
studies - Overall assessment, formalization and
consolidation of SW systems and processes - Work towards Computing MoU in the collaboration