+ All Categories
Home > Documents > Enabling Pervasive Grids The OGF GIN Effort

Enabling Pervasive Grids The OGF GIN Effort

Date post: 02-Feb-2016
Category:
Upload: clio
View: 35 times
Download: 0 times
Share this document with a friend
Description:
Enabling Pervasive Grids The OGF GIN Effort. Erwin Laure GIN-CG co-chair, EGEE Technical Director [email protected]. Worldwide Grid Infrastructures. G I N. APAC DEISA EGEE Naregi NDGF NGS OSG Pragma Teragrid. The Opportunity. - PowerPoint PPT Presentation
14
© 2006 Open Grid Forum Enabling Pervasive Grids The OGF GIN Effort Erwin Laure GIN-CG co-chair, EGEE Technical Director [email protected]
Transcript
Page 1: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum

Enabling Pervasive Grids The OGF GIN Effort

Erwin LaureGIN-CG co-chair, EGEE Technical [email protected]

Page 2: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum

Worldwide Grid Infrastructures

APAC

DEISA

EGEE

Naregi

NDGF

NGS

OSG

Pragma

Teragrid

GIN

Page 3: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 3

The Opportunity

• Grid infrastructures moved from testbeds to dependable production services

• Provide unprecedented combination of compute and storage resources on a global level

EGEE Production Grid Infrastructure Steady growth over the lifetime of the project Improved reliability

32%

Page 4: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 4

The Challenge

• Most applications run on one infrastructure• Requires significant investment – Grids are still hard to use• Example: Astrophysics on EGEE

• Planck, Magic

• Some applications, in particular ones based on international collaborations require seamless access to multiple infrastructures• Usage even harder

Page 5: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum

LCG depends on two major science Grid infrastructures (plus regional Grids)

EGEE - Enabling Grids for E-ScienceOSG - US Open Science Grid

Example: High Energy Physics (LCG)

Page 6: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum

Amsterdam, NL

Orsay,FR Garching,DE

Bologna, IT

Jülich, DE

Argonne, IL

Bloomington , INUrbana-

Champaign , ILSan Diego, CA

InvolvedTeraGrid Sites

Involved DEISA Sites

Cross InfrastructureData Transfer via linkedGPFS clusters

Page 7: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum

SRB Data Grid Federation Status Data Grid

Country SRBversion

Demouserggfsdsc

SRB Zone name Storage ResourceLogical Name

I/OMB/sec

APAC Australia 3.4.0-P yes AU StoreDemoResc_AU 3.9 NOAO Chile/US 3.4.2 yes noao-ls-t3-z1 noao-ls-t3-fs ChinaGrid China CGSP-II (software) RNP Brazil 3.4.1-P2 yes GGF-RNP demoResc UERJ Brazil 3.4.1-P2 yes UERJ-HERPGrid demoResc IN2P3 France 3.4.2 yes ccin2p3 LyonFS4 [25.] DEISA Italy 3.4.0-P yes DEISA demo-cineca KEK Japan 3.4.0-P yes KEK-CRC rsr01-ufs 7.4 SARA Netherlands 3.4.0-P yes SARA SaraStore IB New

Zealand3.4.1 yes aucklandZone aucklandResc (0.3)

ASGC Taiwan 3.4.0-P yes TWGrid SDSC-GGF_LRS1 (0.1) NCHC Taiwan 3.4.0-P yes ecogrid ggf-test CCLRC UK 3.4.0-P yes tdmg2zone IB UK 3.4.1 yes avonZone avonResc WunGrid UK 3.3.1 (hardware) SDSC-wun sfs-tape LCDRG US 3.4.2-P2 Yes LCDRG-GGF demoResc Purdue US 3.4.0-P yes Purdue uxResc1 (2.5) Teragrid US 3.4.0-P2 yes SDSC-GGF sfs-disk U Md US 3.4.0-P yes umiacs narasrb02-unix1

Page 8: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 8

Why is it difficult?

• Grid infrastructures use different technologies• And even if same technologies are used they are usually

heavily customized

• Only a few widely adopted standards• gridFTP, X.509 (but used differently!)• Prototypes: BES, JSDL, …

• Production Grids are difficult to change – adopting standards takes time

• Standards need to be stable before adoption

• Strong interactions between infrastructures and application community needed

Page 9: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 9

GIN-CG

• Started at SC’05• Develop pragmatic techniques for

interoperability• Provide feedback to standards groups,

middleware developers, and Grid operators• Initial focus on authentication, information system,

job submission, file based data movement

• Does NOT provide a pervasive infrastructure• Access needs to be negotiated with the respective

infrastructure

Page 10: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 10

Authentication

• Example AUP

• X.509 certificates and VOMS based VO management• Individual techniques (e.g. gridmapfile) derived from VOMS

information• Some Grids, in particular Supercomputer Grids, require

additional authentication

• Recommendation:• Set up a VOMS server for the VO

Page 11: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 11

Information

• Common top level database for service discovery• Allows to find service endpoints for all Grids• Available in different schemas used

• GLUE schema now being standardized by OGF-WG

• Recommendation• Use this top level database for service discovery

Page 12: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 12

Data Transfer

• gridFTP commonly supported• SRM and SRB based Grids

• Interoperability is being worked on

• Recommendation• Decide on what technology to use• Use gridFTP for data transfer

Page 13: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 13

Job Submission

• Currently many different systems used, event GT2 based systems differ

• BES prototypes demonstrated at SC’06• No adoption path available, yet

• Recommendation• Usage of portals• Usage of VO specific submission systems

Page 14: Enabling Pervasive Grids  The OGF GIN Effort

© 2006 Open Grid Forum 14

Summary

• Pervasive access to Grid infrastructures is possible• Difficult and labor intensive

• Strong interactions between infrastructures and community needed

• GIN can help in providing the contacts and initial recommendations• Findings should be fed back into GIN


Recommended