LHC-B Computing

1
LHC-B Computing

• Computing Tasks
• Computing Model
• Software Strategy
• Objectives
• Technology
• People
• Project Plan for LHC-B Computing
• Project organization
• Development phases and major milestones

2
LHC-B Computing Group

• Pavel Binko
• Olivier Callot
• Lola Camacho
• Marco Cattaneo
• Jean-Pierre Dufey
• Yuri Ermoline
• Markus Frank
• Frank Harris
• John Harvey
• Eric van Herwijnen
• Irakli Mandjavidze
• Pere Mato
• Hans Muller
• Andreu Pacheco
• Florence Ranjard
• Wolfgang Tejessy
• Andrei Tsaregorodtsev

3
Introduction

• New software for a new experiment
• intense use of software in triggering (Levels
  1,2, and 3)
• many orders of magnitude more data and CPU power
• Need better quality Control
• better designs, better control procedures, more
  manpower intensive
• Resources scarce
• improve efficiency - use engineering practices
  and modern technologies
• strategy must encourage all members of
  collaboration to participate
• Need experienced people
• LHC-B staff active in current experiments - big
  advantage
• Need to close culture gap
• Long lead times
• Important to have a well-defined strategy and
  investment in project management and planning

4
Software StrategyObjectives

• Reuse
• put effort into building high quality components
• become economic by extracting more use out of
  components
• Architecture-driven approach
• identify basic building blocks - 3 levels
  patterns, components, frameworks
• reuse spans across application domains
• event display, geometry DB, data storage models,
  controls, GUI
• Common Infrastructure
• standardise on a set of working principles,
  process, development environment
• minimise diversity of hardware and software
• Commercial software
• Licensing model to minimise cost and ensure
  collaboration-wide availability
• Evolution
• development and maintenance span 20 years
• plan for changes in needs and technology

5
Software StrategyTechnology

• Why Object Technologies (OTs) ?
• Evolved from sound principles of software
  engineering
• Young staff with some training in computing
  familiar with OTs
• Technical prerequisites are in place
• development environments (languages etc.), ODBMS,
  communication standards, frameworks
• Widespread use and support by computing industry
• Adoption by HEP collaborations (BaBar, STAR,
  LHC,)
• Object modeling leads naturally to reusability
• encapsulation, design patterns, component-ware,
  frameworks

6
Software StrategyTechnology - Languages
7
Software StrategyTechnology - Distributed
Computing
OMG
Application Objects
Common Facilities
Object Services

• Different standards
• CORBA, IDL (OMG - ODMG)
• DCOM, OLE (Microsoft)
• Javabeans / RMI (Javasoft)

8
Current LHC components
9
Manpower - Profiles

• Architects
• experienced in design of large software systems,
  familiar with standards.
• deliver architectural design documents and
  frameworks
• Software Engineers
• develop basic building blocks , apply quality
  control
• require formal training in software engineering
  practices
• deliver reference documents and user guides
• Software Toolsmiths
• select and maintain software tools used to build
  our software
• install software tools and give support to
  programmers in their use
• manage the codebase and production processing on
  real and simulated data keeping track of all LHCB
  datasets
• Application Programmers
• must be familiar with software development
  environment and follow quality procedures when
  appropriate
• trained in use of supported languages, design
  methods and tools

10
Migration Strategy

• Harvest knowledge in legacy systems - dump legacy
  thinking
• LHC-B has limited software production activity
  now - this will change
• Longer we delay the more difficult it will be to
  make the transition
• Hands-on experience best way to learn
• Need a framework equipped with basic tools before
  people can start
• Intensive training can start once this framework
  is ready
• Need mentors and knowledge transfer between staff
• Technology still evolving and must be tracked -
  choices at appropriate times
• Benefit from support groups - news services,
  library, seminars, in-house training, support for
  engineering tools, joint projects

11
Other issues addressed in report

• Software development process
• Documentation
• Collaborative Computing
• Training

12
Organization of Computing Projects
Steering Group
planning
architecture

coordination



budget
ON-LINE
OFF-LINE
Analysis
Computing
Recon-
DAQ
DCS
OPS
Simul- ation
SEG
Facilities
struction

• Tools
• methods

Event

Level 2

Control

Framework

Framework

DCS

Computing
Builder

Level 3
room

Tools

Tools

LHC
Model

SDE

Interfaces

Prompt

Production

Safety

CPU farms

CASE

Crates

Full

Desktop

Code

DAQware

Storage
management

Network

Quality
Control
Re-usable Components

Event Display

Document.

Frameworks

Data Quality

Training

CORBA,DCOM

Bookkeeping

Licenses

OS

• Librarian
• Collab
• Tools

Data management (DBMS)

Controls

Event Storage (HPSS)

GEANT4

Graphical User Interfaces

CLHEP

Histograms
13
Links to Sub-detector Groups
Application Project (e.g. DAQ)
RICH Computing Team

• Project Leader
• Vertex
• RICH
• Inner Tracker
• Outer Tracker
• ECAL
• HCAL
• MUON
• Trigger L0
• Trigger L1
• Trigger L2/L3

14
Development Phases

• Identify phases by boundaries which correspond to
  major decisions
• Plan for technology change
• Move to OTs which will have major impact
• FORTRAN to C
• Important to prototype
• gain experience
• study behaviour and performance
• After prototype phase take final decisions on
  technology
• Boundary of prototype and production phase should
  have less major impact
• design review, C to JAVA?

15
Software Planning for Transition Phase (1)
16
Software Planning for Transition Phase (2)
17
Global Software Planning Showing Main Development
Phases
18
Global Planning For DAQ System
19
Summary

• Immediate challenge is a change of software
  paradigm
• Plan to make transition as painless as possible
• Relatively good position since we start later and
  have less investment in legacy code
• We have a kernel of experienced people - to be
  complemented by outside help
• Long term strategy involves a prototyping
  (learning) phase and a production phase
• Organise all computing as one project
• minimise barriers between on-line and off-line
• maximise reuse of software components

20
Software StrategyArchitecture-driven development

• Patterns
• well engineered systems are full of patterns
• means of leveraging expertise of experienced
  architects
• provide a common vocabulary and understanding for
  design principles
• Components
• systems constructed out of re-usable tested
  components, with known characteristics and
  well-defined interfaces
• take longer to build, cost is higher
• often can be commercially obtained
• Frameworks
• set of pre-fabricated building blocks used,
  extended or customised for a particular
  application.
• half code in a typical application dedicated to
  routine operations, that can be built once and
  canned for reuseframeworks give high level of
  reuse
• e.g. framework that provides histogramming,
  statistical analysis, visualisation can be used
  for interactive analysis, data quality
  monitoring, .