Towards Production Grids in Greenfield Regions Dr. Ognjen Prnjat European and Regional Grid Management GRNET - Greek Research & Technology Network .

Post on 04-Jan-2016

225 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

Transcript

Towards Production Grids Towards Production Grids in Greenfield Regionsin Greenfield Regions

Dr. Ognjen PrnjatDr. Ognjen Prnjat

European and Regional Grid ManagementEuropean and Regional Grid ManagementGRNET - Greek Research & Technology NetworkGRNET - Greek Research & Technology Network

http://www.grnet.grhttp://www.grnet.gr

2

GRNET missionGRNET mission To provide high-quality

international and national networking and grid services to the Greek academic & research institutions, to the public and private sector so as to support their research and educational activities

HellasGrid: 6 major clusters with a total of approx 800 CPUs and 90 ΤΒ storage capacity (30 SAN, 60 tape)

Leased lambda 2,5 Gbps PoS

Athens MAN (2,5 Gbps PoS)

Dark Fibre (not yet lit)

Patra

Larissa

Heraclion

Syros

Athens

Chania

Rethymnon

Xanthi

Thessaloniki

Ioannina

Leased lambda 1,25 Gbps Gigabit Ethernet

HG-06-AUTH

HG-04-IASA

HG-05-NDC

HG-03- CTI-CEID

HG-01-GRNET Isabella @ Demokritos

HG-02- ICS-FORTH

Hellasgrid Grid Node

3

GRNET operational Grid GRNET operational Grid activitiesactivities

HellasGrid infrastructure and operations EGEE: Federation representative and Regional

Operations Centre for South-East Europe SEE-GRID: “eInfrastructure inclusion” into SEE EUMEDGRID and EUCHINAGRID: Grid

expansion in Mediterranean and China

All greenfield regions!

4

Greenfield regions: challengesGreenfield regions: challenges

Only limited hardware and small clusters available

No previous operational GRID know-how, only research and pilots

Limited operator and user base

5

Key: coordinated actionsKey: coordinated actions

Training the administrators and users Setting up pilot Grid clusters Developing organizational and operational

models Testing and supporting pilot applications

6

Hierarchy: EGEE-SEE ROCHierarchy: EGEE-SEE ROC

7

Organizational modelOrganizational model Clear relations and modus operandi Central coordination by top-level

ROC Responsibility for oversight of

infrastructure status, deployment, and operations

Country-level operations centers and managers

Responsibility for local cluster operational and user support

Distribution of management services across country operations centers

To spread the know-how and ensure joint responsibility and control

Clear policies for site registration and certification

CENTRAL ROC

Country 1 Operations Centre

Country N Operations Centre

8

EGEE-SEE infrastructureEGEE-SEE infrastructure

• Large expansions coming up, reaching ~1500 CPUs, 40 TB SAN and 60TB on Tape

9

Grid servicesGrid services

Catch-all CA: crucial for regional coverage Catch-all VO as unifying VO for

experimentation; VOMS Core services (information services,

workload management, file catalogues, etc)

10

Catch-all Certification AuthorityCatch-all Certification Authority

Need of trust and security -> National Certification Authority

At European Level need to cooperate -> EUGridPMA

SEE Catch-all CA providing certificates to countries without CA

Regional partners are assisted to develop their own Grid CA and apply for membership in PMA

11

Grid management servicesGrid management services

Monitoring EGEE GridIce, Gstat, customized SFTs; local Ganglia

mon, smokeping for network mon Helpdesk

One-or-zero Integrated with EGEE GGUS

User registration Regional and local portals in different languages Installation and certification testbed

12

HelpdeskHelpdesk

Response to operational tickets

•Dealt with 250 tickets in 2005 •1/3 of the originated from EGEE CIC-on-Duty•¼ is from EGEE-wide users• The rest local i.e. ~ 1/2

13

CPU usage in 2005CPU usage in 2005

•TOTAL 552340 CPU Hrs; 23014 CPU Days; 63 CPU Years

14

CPU usage in 2005CPU usage in 2005

15

Usage per VOUsage per VO

16

Local users: VO distributionLocal users: VO distribution

17

SEE Grid-enabled eInftastructure Established at least one fully operational and certified

grid site in all participating SEE countries Two additional Grid applications developed by SEE-

GRID partners deployed in the regional infrastructure: Volumetric Image Visualization Environment (VIVE) for

medical images and other static or time-dependent scalar and vector 3D fields

Search Engine for South-East Europe (SE4SEE) for Grid-aided web-crawling & data indexing

Establishing NGIs more at www.see-grid.org

SEEREN

SEEREN

SEEGRID: the “eInfrastructure SEEGRID: the “eInfrastructure inclusion” into SEEinclusion” into SEE

18

SEE-GRID snapshotSEE-GRID snapshot

19

SEE-GRID vs EGEE-SEESEE-GRID vs EGEE-SEE

EGEE-SEE Grid SEE-GRID Grid

EGEE-wide VOs

EGEE-SEE core services

Strategy: as sites mature EGEE production, adopt operational procedures

Have clear/quantified target/criteria for migrating to EGEE

SEE-GRID VOs

SEE-GRID core services

20

Operational recommendationsOperational recommendations

Adopt best practices of experienced regions, but avoid just copy+paste solutions

Setup a scalable organizational structure Aim to have possibility of stand-alone

operations, independent on related federated Grids and projects

Distribute Grid and management services to spread the know-how and ensure joint responsibility and control

21

Looking ahead: EUMED & Looking ahead: EUMED & EUCHINAGRIDEUCHINAGRID

Creation of regional eScience community

Setting up pilot Grid in the region

Supporting local applications

Foster the creation of a intercontinental eScience community

Training people Supporting existing & new applications

Support interoperable infrastructure for grid operations between Europe and China

22

More recommendations- SEEGRID More recommendations- SEEGRID policy workshoppolicy workshop

There can be no (product-quality, high-performing) grid without a (product-quality, high-capacity, reliable) network: NRENs, GEANT + extensions

Study both “best-practices” and “bad-practices” followed in already grid-established regions and countries.

Respect available infrastructure and resources in the new regions where e-Infrastructures are to be expanded by making provisions for alternative technical roadmaps to fit the area.

Establishing a “web of trust” that brings human capital together is a key priority for expanding eInfrastructures to greenfield regions.

To achieve sustainability, it is necessary to carry out several coordinated and complementary actions: European funds, donations, political support!

23

Greenfield regions: what is it Greenfield regions: what is it about at the end?about at the end?

Increasing the retention of talented scientists in the region

Pursuing joint R&D efforts among countries Making available the benefits of the Information

Society for all citizens Easing the digital divide between the region and

the countries at other side of spectrum Improvement of regional competitiveness in all

market sectors Regional political stability and cohesiveness

24

Thanks to:Thanks to:

All members of EGEE-SEE ROC, SEEGRID project, and new EUMEDGRID and EUCHINAGRID colleagues!

top related