Folie 1

1
Core D-Grid Infrastructure
AstroGrid-D Meeting
ARI Heidelberg
24./25.07.2006
2
Topics of the Infrastructure Project DGI

• FG 1 D-Grid base software (Univ. Dortmund)
• Globus, Unicore, LCG, GridSphere, GAT
• Large Storage, Data-Interface
• VO, Management, Tools
• FG 2 Installation and operation of the
  infrastructure
• Installation and integration of the Core D-Grid
• Operation and support infrastructure
• Ressourcen description and benchmark
• Monitoring, accounting, billing
• FG 3 Network and security (DFN)
• Transport protocols, VPN
• AAI, firewalls, CERT
• FG 4 D-Grid project management (FZK)
• Project management
• Sustainable buisiness models

(FZJ)
(Uni Hannover)
3
Partners of the Core D-Grid
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/sIntel IA32 and IA 64 Cluster, IBM p690, SunFire 80 55 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.
4
Integration of the resources into the Core D-Grid
User/Application
Application development and user access
GAT API
Plug-In
GridSphere
UNICORE
Nutzer
higherGridFunctions
SchedulingWorkflow Management
Monitoring
LCG/gLite
Datenmanagement
Baseservices
AccountingBilling User/VO-Mngt
Globus
Security
Network-infrastructure
distributed compute Resources
Data/Software
distributeddata archives
5
Core D-Grid state of the work
Operational middleware installations

• LCG/gLite
• FZK
• Globus
• LRZ, FZK, TU-Dresden, (FZJ)
• UNICORE
• FZJ, RWTH, LRZ, RZG, TU-Dresden, PC2
• dCache
• FZJ

Operation of the infrastructure (under
development)

• Registration authorities
• VO and user management
• User support (helpdesk)
• Point of Information

6
LCG Services in Karlsruhe
other sites
external UI
Internet
ports2170,7772 9000-9002, 2811, 20000-25000
ports2119,2811 20000-25000
DESY
one port per VO dgtest15000
ports5010,8085
hostdgrid-rls.fzk.de
hostdgrid-rb.fzk.de
ports 2811 (GridFTP), 20000-25000 (Globus Port
Range - GPR)
Information Index
Resource Broker
File Catalog
hostdgrid-voms.fzk.de
hostdgrid-se.fzk.de
Storage Element
local disk
hostdgrid-ce.fzk.de
User Interface
Computing Element
Computing Element
ssh
User
hostdgrid-ui.fzk.de
7
Globus configuration
GSI grid-proxy-init
Globus client
GridFTP globus-url-copy
GRAM globus-job-run globusrun-ws
MDS4 wsrf-query
Firewalls
Compute resource ZIH
Compute resource FZK
Compute resource LRZ
8
UNICORE configuration
Unicore clients
usitedgrid
Firewalls
Port 4009
Unicore Gateway
Port 4009
Port 8181
NJS
NJS
NJS


FZJ/JUMP
TSI
Compute resource JUMP
9
Status of the UNICORE installations
Vsite CPU Peak performance TFlops
FZJ_JUMP 32 Power 4 (IBM p690) 0.218
RZ_RWTH-AACHEN 48 Sparc / Opteron 0.23
ZIH_A3K 192 Itanium (SGI Altix) 1.1
ZIH_O3K 128 MIPS (SGI Origin) 0.102
ZIH_O3K2 64 MIPS (SGI Origin) 0.051
lrz-itaniumcluster 134 Itanium 0.86
lrz-migrationscluster 32 Itanium (SGI Altix) 0.41
lrz-pentiumcluster 138 Pentium 0.80
rzg-sp5 20 Power 5 (IBM p575) 0.152
pc2-arminiuscluster 416 Xeon / Opteron 2.6
10
Storage Element in Jülich
VOILDG, DGI
Firewall
gridftp
https, etc
DMZ
domain.zam.kfa-juelich.de
hostzam213
hostzam420
hostzam214
GPFSCommon Filesystem
SRM/Admin Server

UI
hostzam452
Blue Gene/L JUBL
dCache
Supercomputer JUMP
TSMStorage Backend
11
Core D-Grid access to Compute Resources

• Public key infrastructure
• Grid certificates by DFN or FZK
• Registration authorities at partner sites
• FZJ/ZAM, FZK/IWR, LRZ, RWTH/RZ, RZG/MPG, PC2,
  TUD/ZIH
• RAs of DFN
• RAs of FZK
• Registration
• Multi-community ressources
• 3-level hierarchy
• Community ? VO ? VO member
• Interface to VOMS

12
Core D-Grid access to Compute Resources
13
Core D-Grid access to Compute Resources
14
Core D-Grid access to Compute Resources
Community VO VO-Speaker Resource requested approved Status
Kern-D-Grid administrator W. Gürich Jump (FZJ) 555 0 requested
Astro-Grid-D sun S. Sonne SunFire (AC) 20 15 approved
Kern-D-Grid trial Ch. Dohmen Opteron (FZK) 300 0 rejected
Medi-Grid mri M. Müller Xeon (FZK) 266 266 approved







15
Installation of a Support Infrastructure
Communities
DGI
D-Grid User Support common by FZK and FZJ
common trouble ticket system or compatible system
s
DGUS
Operation mngt.
User support
16
D-Grid User Support
User requests
Community Consulting Center
D-Grid User Support
Resource-Provider
Service / middleware administrator
17
D-Grid User Support
18
D-Grid User Support

• D-Grid User Support
• dgus.d-grid.de
• Generation of trouble tickets
• User support e-mail addresses of partners
• dgrid-support_at_fz-juelich.de
• dgrid-support_at_fzk.de
• dgrid-support_at_lrz.de
• dgrid-support_at_rzg.mpg.de
• dgrid-support_at_rz.rwth-aachen.de
• dgrid-support_at_tu-dresden.de
• dgrid-support_at_uni-paderborn.de