The first year of the Estonian Grid

1
The first yearof the Estonian Grid

• Andi Hektor, Mario Kadastik, Martti Raidal --
  NICPB
• Lauri Anton, Hardi Teder -- EENet
• Konstantin Skaburskas -- Tartu Univ.
• Vahur Kotkas -- Tallinn Univ. of Tech.
• andi.hektor_at_cern.ch
• NGN Workshop in Linöping
• Oct 20, 2004

2
Birth of the Estonian Grid (EG)
Jan 29, 2004 the CA (and the phone of the
RA), the country GIIS, the first CE
3
Brief history of the EG

• Jan '04 first components of the EG EG CA,
  country level GIIS and first test clusters
• Jan '04 NorduGrid technical meeting in Tallinn
• Feb '04 Centre of High Energy Physics and
  Computational Sciences at NICPB
• Mar '04 some first multiprocessor clusters join
  the EG
• Apr '04 CP/CPA draft for the EG CA
• May '04 first scientific software ported to the
  EG
• May '04 Grid laboratory at the Tartu University
• Jun '04 establishment of the technical
  coordination group of the EG
• Jun-Jul '04 first massive scientific
  calculations at the EG
• Okt '04 EG technical support group publishes the
  first article about the first scientific results
  produced with the EG

4
Brief history of the EG

• Oct 5-7 04 The first Baltic Grid meeting

5
Current situation of the EG

• Choice of middleware NorduGrid ARC
• EG CA is trusted by EUGridPMA and NorduGrid
• 62 CPUs in the EG (Pentium4 2.23.0 GHz)
• The country GIIS registers to NorduGrids top
  level GIISes

?
?
6
Technical support for the EG users

• Technical support and coordination group, 6
  persons at the different institutions
• Support list grid-tech_at_lists.eenet.ee
• Web http//grid.eenet.ee

7
Know-how centres

• Estonian CERN CMS group at NICPB Porting CMS
  software to NorduGrid CERN Tier 2
  plans http//www.nicpb.ee/centre
• UT Lab of Grid technology DOUG
  software Commercial applications
• UT Theoretical chemistry UNICORE,
  OpenMolGrid http//www.openmolgrid.org
• TUT Lab of cluster technology Environmental
  science Hardware systems
• Plans bioinformatics, medicalapplications,
  nuclear safety

8
Structure of the EG
NICPB
UT
TUT
EENet
Country-level GIIS
EG CA
Technical support group
Technical support group
Lectures, seminars, user support
Grid Laboratory
Centre for HEP comp. sciences
Cluster Laboratory
EAS
CERN CMS software, Geant4, etc
Software for nonlinear dynamics, etc
DOUG software, ID-card, etc
Commercial applications
9
Political structure and cost of the EG

• The political structure of the EG - the steering
  committee of the EG is establishing - EENet,
  NICPB, TUT, UT
• The cost of the EG - 0 euros in this year -
  based volunteers in this year - 4-6 persons are
  involved, not full-time - the budget of the next
  year is open

10
Some excellent possibilities of the EG

• Electronic ID-card infrastructure (600 000
  ID-cards), it is easy to use for the Grid
• Local experiences with E-money and rental
  software
• Good connections to NorduGrid

11
Support project for the commercial applications
Coordinator The Lab of Grid Technology at the UT
Companies

• AS EGeen analyse of gene information
• OÜ Nukufilm animations and rendering
• OÜ Vendomar testing of the applications
• AS Sertifitseerimiskeskus ID-card and its
  applications
• AS Silmet nuclear and radiation safety
• AS Tallinna Diagnostikakeskus medical
  applications
• Other Centre of Radiation Safety, the Ministry
  of Defence

Collaboration

• Estonian Grid, NorduGrid
• NICPB
• Institute of Physics at the UT

12
Some applications Geant4 (1/3)

• Medical applications
• Nuclear safety
• Materials for the nuclear industry

13
Some applications Geant4 (2/3)

• Mobile gamma spectrometry
• Need for modelling Monte Carlo (Geant4)
• Computational power NorduGridi, EG
• Collaboration NICPB and UT in Estonia STUK, HIP
  and MATINE in Finland

14
Some applications Biotech (3/3)

• EGeen, Estonian Gene Foundation
• Personalized medicine
• Analyse of the gene information on the Grid

15
First scientific results from the EG nuclear
safety calculations using Geant4

• The total time of the computations was 417 CPU
  days. The speed of the CPUs used were between 2.4
  GHz and 3.06 GHz (IntelPentium 4).
• In total there were 17 failed jobs (2.6) during
  the computations, probably due to random hardware
  errors. There were 16 post-processing errors
  (2.5) due to the instability of the
  hardware/software caused by the external factors
  blackouts of electric grid, overheating, etc.

16
Thank you!