US-CMS User Facility and Regional Center at Fermilab

1
US-CMS User Facilityand Regional Centerat
Fermilab

• Matthias Kasemann
• FNAL

2
Outline

• User Facility
• Goals Schedule Organization
• US-CMS Regional Center at Fermilab
• Hardware and Networking
• Data required
• Support Functions
• Size of Regional Center
• Cost and Personnel Profile
• Summary

3
User Facility Goals

• Scope and Size support US-CMS collaboration
  (20 of full CMS).
• Provide enabling infrastructure of software and
  computing to allow US physicists to fully
  participate in physics program of CMS.
• Support the development and data analysis
  activities of US-CMS
• acquire, develop, install, integrate, commission
  and operate hardware and software.
• This will include a major Tier1 regional
  computing center at Fermilab to support US
  physicists working on CMS.

4
US_CMS RC1 at FNALSchedule Organization

• 1999 - 2003 RD Phase
• 2003 - 2005 Implementation Phase
• 2006 onwards Operations Phase
• All Phases of the US-CMS RC1 will be managed and
  operated within the FNAL Computing Division.
• FNAL-CD is collecting very qualified experience
  from Tevatron RunII preparations and operations.
• This will put us in a unique position to support
  collaborations of
• 4-500 CMS scientists doing pp physics.

5
Fermilab Run II Computing and Software

• Run II Parameters for DØ
• Trigger rate 50 Hz (LHC / 2)
• Raw data event size 250 kB (LHC / 4)
• Data collection 6 x 108 evts/ yr. (LHC
  / 1.6)
• Summary event size 150 kB (LHC x 1.5)
• Physics sumry evt size 10 kB (LHC)
• Total dataset size 300 TB/yr. (LHC / 3)
• Bottom line Computing project O (Run I x 20)
• 
  O (LHC / 2-3)
• This will be accomplished with resources
  available in 2000.

6
US-CMS Regional Center at FNALHardware and
Networking

• Implement the main US-CMS regional center at
  Fermilab
• include substantial CPU for analysis,
  reprocessing and simulation
• data storage
• data access facilities.
• Networks and networking will play a key role in
  this distributed computing model
• Deliver data to other US-CMS institutions through
  high-speed networks.
• High bandwidth network connection to CERN.
• Technology and policy in this area is in a state
  of flux. We have to track technology and
  participate in prototyping work.

7
Data required at US-CMS RC1

• Event data available at regional center
• event samples for testing and code development,
• a collection of very interesting events
• the full set of the Analysis Object data (AOD)
• about 10 of the raw and reconstructed data (ESD)
• all available event meta data
• Non-event data
• detector databases for calibration, monitoring,
  geometry, cabling
• data related databases production log, run
  conditions, trigger setups

8
US-CMS Regional Center at FNALSupport Functions

• Include user support personnel
• training, documentation, code distribution, ...
• Personnel to manage licenses and license
  acquisition.
• Contract for needed services.
• Responsibility and personnel to develop or
  acquire any software that is required to carry
  out its production and operation activities.
• Provide support for many development activities
  during the detector construction period before
  data taking begins.
• Provide ongoing support during operations phase
  of the experiment.

9
RD Phase activities

• Participate in RD to prove the concept of LHC
  regional centers
• MONARC testbed
• Object database testbeds
• Prototype regional center by 2002
• Networking testbeds
• In RD phase and beyond
• Provide CMS user support
• documentation
• code management and distribution
• training
• user help desk

10
Size of the Tier 1 Regional Center

• The base for these estimates is a set of figures
  for CMS offline computing at CERN that was made
  available by CMS in mid-1998.
• CPU (1 TIPS 106 MIPS ? 25k SpecInt95)
• in 2005 3.6 TIPS
• 2006 on 1.2 TIPS new 1.2 TIPS replacement
• Disk storage
• in 2005 108 TB
• 2006 on 40 TB new 27 TB replacement
• Serial storage
• in 2005 1 robot 0.4 PB
• 2006 on 0.2 PB new robot every 3 years

11
Hardware cost estimation

• Costs de-escalated according to technology-cost
  development expectations.
• Actual costs for CPU and disk are based on our
  Run II experience
• CPU 3.5M/TIPS in 2003 (extrapolated from 1999)
• disk 26.90/GB in 2003 (extrapolated from 1999)
• The cost reflect the use of expensive SMP
  machines for analysis in Run II.
• Cost for serial storage also based on Run II
• 1.1M per robot plus 0.5M/PB for media

12
Regional Center ScheduleImplementation and
Operation phase

• To distribute cost over 2003, 2004 and 2005 and
  profit from price/performance evolution we plan
  to acquire 10 of disk and CPU in 2003, 30 in
  2004, 60 in 2005.
• Integrated capacity installed per year
• 2003 2004 2005 2006 CMS operation
• CPU 0.4 1.44 3.6 4.8 TIPS
• disk 11 43 108 148 GB
• tape 0.03 0.05 0.4 0.6 PB
• robots 1 0.33 robots

13
Cost and Personnel Profile
14
Cost Summary

• Total costs through 2005
• hardware 10.6 M
• networking in US 2.1 M
• licenses 0.1 M / year
• Operating phase
• 2006 onward 3.9 M hardware 35 people
• Note
• This hardware costs are based on what it is
  costing to serve 400-500 people in CDF and DØ
  under the assumption of exponential
  price/performance evolutions (9.1M for CDF or
  D0).
• Numbers are conservative (SMP-CPU,)

15
US-CMS User FacilitySummary

• The User Facility has to provide US-CMS
  scientists with a competitive infrastructure to
  fully participate in the science program of CMS.
• FNAL with the experience of Tevatron experiments
  is well suited to host the US-CMS major Tier 1
  Regional Center.
• Setup and support of the Regional Center will
  happen during 2003 - 2005. Initial hardware cost
  estimations (w.o. networking) are 10.6M 35
  FTEs (by 2005).
• Operations cost estimates are 2.7M/year for
  hardware (w.o. networking) and about 35 people.
• Networking cost will be substantial.