1 Wolfgang Gentzsch, D-Grid September, 2006
GridKa Karlsruhe, September 11 – 15, 2006
Building the German D-Grid *)
Wolfgang Gentzsch
D-Grid, RENCI, Duke, OGF, PCAST
*) funded by the German Ministry for Education and Research
German e-Science Initiative 2003 Agenda 2006: The German e-Science - - PowerPoint PPT Presentation
GridKa Karlsruhe, September 11 15, 2006 Building the German D-Grid *) Wolfgang Gentzsch D-Grid, RENCI, Duke, OGF, PCAST *) funded by the German Ministry for Education and Research 1 September, 2006 Wolfgang Gentzsch, D-Grid German
1 Wolfgang Gentzsch, D-Grid September, 2006
GridKa Karlsruhe, September 11 – 15, 2006
Wolfgang Gentzsch
D-Grid, RENCI, Duke, OGF, PCAST
*) funded by the German Ministry for Education and Research
2 Wolfgang Gentzsch, D-Grid September, 2006
3 Wolfgang Gentzsch, D-Grid September, 2006
“distributed computing”, like e.g. Unicore,…
Many German research centers are part of European grid projects, like DataGrid, GridLab, EuroGrid, CrossGrid, EGEE, GridCoord, CoreGrid, UniGrids, NextGrid,…
4 Wolfgang Gentzsch, D-Grid September, 2006
5
Wolfgang Gentzsch, D-Grid September, 2006
6 Wolfgang Gentzsch, D-Grid September, 2006
Information and Knowledge Management
Grid specific Developments
Application CG Middle- ware
Grid specific Development
Application CG Middle- ware
Courtesy Dr. Krahl PT/ BMBF
7 Wolfgang Gentzsch, D-Grid September, 2006
8 Wolfgang Gentzsch, D-Grid September, 2006
9 Wolfgang Gentzsch, D-Grid September, 2006
10 Wolfgang Gentzsch, D-Grid September, 2006
Nutzer Application Development and User Access GAT API
Data/ Software
Resources in D-Grid High-level Grid Services Basic Grid Services
Distributed Data Archive
User
Network Infrastructur
LCG/gLite Globus 4.0.1 Accounting Billing User/VO-Mngt
Scheduling Workflow Management
Data management Security Plug-In UNICORE
Distributed Compute Resources
GridSphere Monitoring
Wolfgang Gentzsch, D-Grid September, 2006
12 Wolfgang Gentzsch, D-Grid September, 2006
13 Wolfgang Gentzsch, D-Grid September, 2006
PC² RRZN TUD RZG LRZ RWTH FZJ FZK FHG/ ITWM Uni-KA Site Resource Amount
FZJ/ZAM IBM Supercomputer with 8,5 TFlops STK data robot system with 2,8 PByte 32 CPUs 300 TByte FZK/IWR 8 nodes Opteron 2x2.2 GHz 8 processors of a system NEC SX-5 1 p630 with 4 processors 1 SX-6i to do tests 2 nodes Opteron 2x2.2 GHz to do tests 100% 50% 50% 50% 50% LRZ SGI high performance system with 20 TFlop/s Intel IA32 and IA 64 Cluster, IBM p690, SunFire 80 5% 5% 5% MPI/RZG IBM supercomputer with 4,5 TFlops, PC cluster with 2 TFlops Data robot system with 8 PByte 32 CPUs 400 TByte PC² Cluster of 400 Xeon 64 Bit processors, high performance visualization and FPGAs 10% RWTH/RZ 2 SunFire 6900 with 24 UltraSPARC IV each 100% TU- Dresden/ZIH SGI O2K(56 proc)/O3K(192 proc.) : T3E (64 proc): PC cluster with 30 processors, end off 2005: new system with 1000 proc. 10% 20% 20% 2% Uni-H/RRZN PC-Cluster mit 64 CPUs assoc. Uni-KA PC-Pool assoc. FHG/ITWM assoc.
14 Wolfgang Gentzsch, D-Grid September, 2006
5.3 Mio Euro Investment in D-Grid in October 2006
15 Wolfgang Gentzsch, D-Grid September, 2006
16
Joint EGEE / D-Grid Meeting 16
Dienste- Anbieter Dienste- Anbieter Dienste- Anbieter Service provider User support Operation mngt.
Entwickler J Entwickler J Developer Deployment-Team
17
Joint EGEE / D-Grid Meeting 17
18 Wolfgang Gentzsch, D-Grid September, 2006
19 Wolfgang Gentzsch, D-Grid September, 2006
20 Wolfgang Gentzsch, D-Grid September, 2006
21 Wolfgang Gentzsch, D-Grid September, 2006
22 Wolfgang Gentzsch, D-Grid September, 2006
Stepping from Climate (=Atmosphere+Ocean) to Earth System Modelling Earth system model wishlist: Higher spatial and temporal resolution Quality: Improved subsystem models Atmospheric chemistry (ozone, sulfates,..) Bio-geochemistry (Carbon cycle, ecosystem dynamics,..)
Increased Computational demand factor: O(1000 -10000)
23 Wolfgang Gentzsch, D-Grid September, 2006
Crossing rate 40 MHz Event Rates: ~ 109 Hz Max LV1 Trigger 100 kHz Event size ~ 1 Mbyte Readout network 1 Terabit/ s Filter Farm ~ 107 Si2K Trigger levels 2 Online rejection 99.9997% (100 Hz from 50 MHz) System dead time ~ % Event Selection: ~1/1013 Crossing rate 40 MHz Event Rates: ~ 109 Hz Max LV1 Trigger 100 kHz Event size ~ 1 Mbyte Readout network 1 Terabit/ s Filter Farm ~ 107 Si2K Trigger levels 2 Online rejection 99.9997% (100 Hz from 50 MHz) System dead time ~ % Event Selection: ~1/1013
Event rate
“Discovery” rate
Luminosity Low 2x1033 cm-2 s-1 High 1034 cm-2 s-1 Data analysis: ~ 1PB/ year Level 1 Trigger Rate to tape
Courtesy David Stickland
24 Wolfgang Gentzsch, D-Grid September, 2006
Gridspezifische Entwicklungen
Cooperation and business models
Molding Metal Form ing Fluid Processes Groundw ater Transportation
Knowledge-based support for engineering-specific decision support Support for engineering- specific Workflows Distributed simulations-based product & process
Fluid-Structur/ Magneto-Hydro- dynam ic I nteraction
Security and trust models Grid-specific developm ents
25 Wolfgang Gentzsch, D-Grid September, 2006
Raw Data
Metadata
Molecule
Homogenization Target data Metadata
Population
Metadata
Patient
Metadata
Illness
Metadata
Organ/Tissue
Metadata
Cell
Search, Find, Select
Access Control
Correlate, Process, Analyze Resulting Data Presentation Final Result
26 Wolfgang Gentzsch, D-Grid September, 2006
27 Wolfgang Gentzsch, D-Grid September, 2006
28 Wolfgang Gentzsch, D-Grid September, 2006
29 Wolfgang Gentzsch, D-Grid September, 2006
Courtesy Dieter Kranzlmueller
30 Wolfgang Gentzsch, D-Grid September, 2006
31 Wolfgang Gentzsch, D-Grid September, 2006
32 Wolfgang Gentzsch, D-Grid September, 2006
33 Wolfgang Gentzsch, D-Grid September, 2006
34 Wolfgang Gentzsch, D-Grid September, 2006