View
221
Download
0
Tags:
Embed Size (px)
Citation preview
1
Kihyeon Cho &Kihyeon Cho &Soonwook Hwang Soonwook Hwang
(KISTI)(KISTI)
Super Belle with FKPPL VO Super Belle with FKPPL VO & AMGA Data Handling& AMGA Data Handling
GoalGoal
Background Collaborative work between KISTI and CC-IN2P3 in the area of
Grid computing under the framework of FKPPL Objective
(short-term) to provide a Grid testbed to the e-Science summer school participants in order to keep drawing their attention to Grid computing and e-Science by allowing them to submit jobs and access data on the Grid
(long-term) to support the other FKPPL projects by providing a production-level Grid testbed for the development and deployment of their applications on the Grid
Target Users FKPPL members 2008 Seoul e-Science summer school Participants
FKPPL TestbedFKPPL Testbed
Service Host Site
UI kenobi.kisti.re.kr KISTI
VOMS palpatine.kisti.re.kr KISTI
WMS/LB snow.kisti.re.kr KISTI
SE ccsrm02.in2p3.fr (0.5TB) CC-IN2P3
hansolo.kisti.re.kr (1.5TB) KISTI
CE cclcgceli03.in2p3.fr (5000 CPU cores) CC-IN2P3
darthvader.kisti.re.kr (100 CPU cores) KISTI
VOMS
WMSCE
SE
UI
CE
SE
FKPPL VO
KISTIIN2P3
LFC WIKI
VO Registration Detail Official VO Name
fkppl.kisti.re.kr
Description VO dedicated to joint research projects of the
FKPPL(France Korea Particle Physics Laboraroty), under a scientific research programme in the fields of high energy physics (notably LHC and ILC) and e-Science including Bioinformatics and related technologies
Information about the VO https://cic.gridops.org/index.php?section=vo
FKPPL VO Usage
Application porting support on FKPPL VO Geant4
Detector Simulation Toolkit Working with National Cancer Center
WISDOM MD part of the WISDOM drug discovery pipeline Working with the WISDOM Team
Support for FKPPL Member Projects
Grid Testbed for e-Science School Seoul e-Science summer school
How to access resources in FKPPL VO Testbed
Get your certificate issued by KISTI CA http://
ca.gridcenter.or.kr/request/certificte_request.php
Join a FKPPL VO membership https://palpatine.kisti.re.kr:8443/voms/fkppl.k
isti.re.kr
Get a user account on the UI node for FKPPL Vo Send an email to the system administrator at
User Support
FKPPL VO Wiki site http://anakin.kisti.re.kr/mediawiki/index.php/FKPPL_VO
User Accounts on UI machine 17 User accounts have been created
FKPPL VO Registration 4 users have been registered as of now
Contact Infomation Soonwook Hwang (KISTI), Dominique
Boutigny (CC-IN2P3) responsible person [email protected], [email protected]
Sunil Ahn (KISTI), Yonny Cardenas (CC-IN2P3) Technical contact person, [email protected], [email protected]
Namgyu Kim Site administrator [email protected]
Sehoon Lee User Support [email protected]
Monday, December 2, 2008Yonny CARDENAS
11
KISTI site
VOMS, WMS, CE+WN*, UI, Wiki* Infrastructure installation in progress ( a cluster with 128 cores has
been purchased)
CC-IN2P3 site
CE+WN, SE, LFC
Configuration
Monday, December 2, 2008Yonny CARDENAS
12
VO Registration procedure
– VO name: fkppl.kisti.re.kr
– VO manager: Sunil Ahn
– Status: Active
Configuration
Monday , December 2 , 2008Yonny CARDENAS
13
Status (Operational Services)
KISTI site: VOMS OK WMS OK CE OK WIKI OK SE OK WN* in Progress
CC-IN2P3 site: SE OK dCache/SRM OK WN OK CE OK LFC OK
Monday, December 2 , 2008Yonny CARDENAS
14
Available Services
Job Submission
Available since October 1, 2008
Resource allocation: 5 millions hours CPU SI2K
CC-IN2P3 Job monitoring
Quality of Service
Operation team
Monday , December 2 , 2008Yonny CARDENAS
15
Available Services
Data storage
dCache SE/SRM
System for storing and retrieving data, distributed among a large number of heterogeneous server nodes.
Implements the SRM v2.2 interface required EGEE/LCG
Resource allocation: 0.5 Terabytes
Monday, December 2, 2008Yonny CARDENAS
16
Available Services
Data storage
AFS (Andrew File System)
Network file system for personal and group files, experiment software, system tools (compilers, libraries , ... )
Indirect use (jobs)
Resource allocation: 2 Gigabytes
Monday, December 2 , 2008Yonny CARDENAS
17
Available Services
Data storage
LFC - LCG File Catalog
Maintains mappings between logical file names (LFN) and SRM file identifiers.
Supports references to SRM files in several storage elements.
Monday, December 2, 2008Yonny CARDENAS
18
Utilisation - Services
Jobs Submission
October
34 jobs for 150 hours CPU SI2K
November 1690 jobs for 48250 hours CPU SI2K
Monday, December 2, 2008Yonny CARDENAS
19
Utilisation - Services
Data Storage
7193 files for 60 G bytes of used space
• 440 G bytes available.
User SupportUser Support
FKPPL VO Wiki site http://anakin.kisti.re.kr/mediawiki/index.php/FKPPL_VO
User Accounts on UI 20 User accounts has been created
FKPPL VO Membership Registration 7 Users have been registered at FKPPL VO membership
FKPPL VO UsageFKPPL VO Usage
Deployment of Geant4 applications on FKPPL VO Detector Simulation Toolkit Working with Jungwook Shin at National Cancer
Center
Grid Interoperability Testbed
Geant4 Application: GTR2_com Geant4 Application: GTR2_com Application name: GTR2_com (G4 app for proton therapy sim s/w by developed by NCC) -> GTR2 : Gantry Treatment Room #2, com: commissioning
(now GTR2 simulation code is under commissioning phase)
-> libraries: Geant4 , root (root.cern.ch) as simulation output library
/user/io/OpenFile root B6_1_1_0.root/GTR2/SNT/type 250/GTR2/SNT/aperture/rectangle open#Geant4 kernel initialize/run/initialize /GTR2/FS/lollipops 9 5/GTR2/SS/select 3/GTR2/RM/track 5/GTR2/RM/angle 80.26 /GTR2/VC/setVxVy cm 14.2 15.2/beam/particle proton/beam/energy E MeV 181.8 1.2 /beam/geometry mm 3 5/beam/emittance G mm 1.5/beam/current n 3000000#SOBP/beam/bcm TR2_B6_1 164 /beam/juseyo/user/io/CloseFile
user macro
output
GTR2_com
GTR2_com 의 input 은 nozzle 의 configuration 이며 , 이 configuration 이 명시된 macro 파일을 읽어서 최종 양성자 빔에 의한 선량분포를 3D-histogram 의 root 파일로 출력
Distribution of the completion time of 1000 Distribution of the completion time of 1000 GTR2_com jobs on FKPPL VO GTR2_com jobs on FKPPL VO
The submission of 1000 GTR2_com jobs was done around 18:05
BC408 M.C study
Purpose : the accurate simulation study will help to design and construct a dosimetry device
utilizing the BC408 scintilator
BC408 M.C on FKPPL resolution : 2mm in X,Y and 1mm
thicknes
From 1 file, ~3.5 hrs on WN of FKPPL
From 589 files, 589 files sucessfully generated among total 99*7 (693)
jobsTrial # Total job Completed Err
1 99 65 34
2 99 66 33
3 99 99 0
4 99 98 1
5 99 84 15
6 99 80 19
7 99 97 2
693 589 104
I immediately submit parametric job after initializing the proxy
Super Belle ComputingSuper Belle Computing
Conveners: T. Hara, T. Kuhr
Distributed Computing (Martin Seviour) Data Handling (Kihyeon Cho) Data Base (Vacant)
Super Belle Data HandlingSuper Belle Data Handling
Data Handling depends on distribution computing. Cloud Computing? Grid farm?
Data Handling Suggestions (2/17) SAM (Sequential Access through Metadata)
CDF by Thomas Kuhr
AMGA KISTI by Soonwook Hwang
EGEE (Enabling Grids for E-SciencE)EGEE (Enabling Grids for E-SciencE)
• the largest multi-disciplinary grid infrastructure in the world
Objectives• Build large-scale, production-quality
grid infrastructure for e-Science
• Available to scientists 24/7
EGEE grid Infrastructure• 300 sites in 50 countries• 80,000 CPU cores• 20 PBytes• 10,000 User
Overview of AMGA (1/2) Overview of AMGA (1/2) Metadata is data about data AMGA provides:
Access to Metadata for files stored on the Grid A simplified general access to relational data stored in database systems.
2004 – the ARDA project evaluated existing Metadata Services from HEP experiments AMI (ATLAS), RefDB (CMS), Alien Metadata Catalogue (ALICE) Similar goals, similar concepts Each designed for a particular application domain
Reuse outside intended domain difficult Several technical limitations: large answers, scalability, speed, lack of flexibility
ARDA proposed an interface for Metadata access on the GRID Based on requirements of LHC experiments But generic - not bound to a particular application domain Designed jointly with the gLite/EGEE team
Overview of AMGA (2/2) Overview of AMGA (2/2)
What is AMGA ? (ARDA Metadata Grid Application) Began as prototype to evaluate the Metadata Interface
Evaluated by community since the beginning: Matured quickly thanks to users feedback
Now part of gLite middleware : EGEE’s gLite 3.1 MW
Requirements from HEP community Millions of files, 6000+ users, 200+ computing centres Mainly (real-only) file metadata Main concerns : scalability, performance, fault-tolerance, Support
for Hierarchical Collection
Requirements from Biomed community Smaller scale than HEP Main concerns : Security
ARDA Project (A Realisation of Distributed Analysis for LHC)ARDA Project (A Realisation of Distributed Analysis for LHC)
Metadata user requirementsMetadata user requirements
I want to store some information about files
In a structured way query a system about those information keep information about jobs
I want my jobs to have read/write access to those information
have easy access to structured data using the grid proxy certificate
NOT use a database
Metadata ConceptsMetadata Concepts Schema (table, think directory)
Has hierarchical name and list of attributes /prod/events
Attributes (columns) Have name and storage type Interface handles all types as strings
Entry (row) Live in a schema, assign values to attributes
Collections A set of entries associated with schema
Query SELECT ... WHERE ... clause in SQL-like or SQL query language
AMGA data organizationAMGA data organization
Relational schema AMGA(hierarchy)
/HOSPITAL//HOSPITAL/
PATIENTS/PATIENTS/ DOCTORS/DOCTORS/
johnjohn george
george
#name sickness age
john malaria 68
george otitis 84sickness otitis
age 84
Attributes
Entries
Schema/Directory
Schema/Directory
TABLE: PATIENTS
#name
PATIENTS
DOCTORS
TABLE: HOSPITAL
Collection
#type
people_group
people_group
Importing existing dataImporting existing data
Suppose that you have the data A reasonable question would be:
Can I use my existing database data?? The answer is YES
Importing data to AMGA Pretty simple Connect a database to AMGA
Execute the import command import table directory
Ready to go!
AMGA Features in One SlideAMGA Features in One Slide Official gLite middleware component for metadata catalogue Metadata is relationally structured
Schema (aka table, think directory)Has hierarchical name and list of attributes /prod/events
Attributes (aka columns)Have name and storage type, Interface handles types as strings
Entry (aka row)Live in a schema, assign values to attributes
Query: SELECT ... WHERE ... clause in SQL Fine grained access control (ACL) support
Table level and entry level Tight integration into the Virtual Organization Management System (VOMS)
X509 Grid certificate Native SQL support in AMGA 1.9
Direct DB access to existing databases on the Grid via SQL OGF WS-DAIR compatible interface support in AMGA 2.0 Uniform interface to Heterogeneous Database backend
Oracle, Postgres SQL, MySQL, etc. Support for Many Programming APIs
Diverse user community requested/provided e.g., C/C++, Java, Python, Perl, PHP
Replication Support Full replication, Partial replication, Federation
Support for “Import of pre-existing databases”