NGI: luxury or Necessity The EELA lessons

1
NGI luxury or Necessity?The EELA lessons

• Bernard M. Marechal (EELA-2 Project Coordinator)
• CETA-CIEMAT UFRJ (Spain Brazil)
• NGI-AR Launching Meeting
• Buenos Aires (Argentina), 22.10.2009

2
e-Science before the DCI
3
e-Science with DCI
4
Short history

• Several meetings (Rio de Janeiro, CERN,
  Itacuruca) in 2004, foreseeing a EULA Grid
  Project supported by EC
• EELA proposal prepared, submitted and approved in
  2005
• EELA worked hard and well during 2 years (2006
  2007)
• EELA-2 prepared submitted in 2007 and approved
  in 2008
• EELA-2 will end on 31 March 2010
• EC support fade out in the future
• LGI / NGIs fade in mandatory to support
  e-Science in LA on the long term

5
Some thoughts

• Forget the war between Grid, Clouds, HPC, .
  And think about DCI (Distributed Computing
  Infrastructures)
• GRID Distributed Supercomputing Facility ()
• HPC Physically Localised Supercomputing Centres
  ()
• GRID more than computing resources (human
  networks, scientific collaborations
• Why a Grid if I have a cluster? And what about
  tomorrow???
• Cores are important but human resources / skills
  are fundamental

6
Some thoughts

• No stress, no fear GRID is not a black hole
  your resources (cluster) remain under your
  control
• GRID means
• I do want to share
• I offer because I need
• NGI bible
• Should be application independent
• Should be MULTI MIDDLEWARE (Middleware is not an
  issue, just a necessity)
• User demand as a seed (highway without cars!)
• Close / strong / interaction / collaboration
  with NRENS and CLARA is a must

7
Some thoughts

• Do not waste your energy with wrong target
• Industry should be attacked (tough job! Fear,
  Clouds, )
• Argentina is mature (thanks, partially, to EELA
  and EELA-2)
• JRU exists
• Motivated people (researchers engineers) are
  active
• Skilled people in GRID are available
• Impressive Governmental support (unique in LA???)
• There is a CA in Argentina (tough job !!!!!!)
• THE DANGER matches effect and unresponsiveness
• ADAPT TO YOUR REALITY BUT NEVER COPY-PASTE
  !...................

8
Argentinas e-Science Collaboration (JRU)
The Joint Research Unit already created with
support from Secyt (now Mincyt)
Members at present LIDI, LINTI ( Fac, of
Informatics UNLP), IFLP (CONICET-UNLP), Univ.
Nac. Rio Cuarto, Univ. Nac. del Sur, UBA, HIBA
(Hospital Italiano Buenos Aires), INNOVARED
8
9
An Example The Argentinean RC
The process of integration of these resources in
the EELA infrastructure has been in parallel with
the learning process and training of the local
people on the GRID technology, but it is finally
ready!

• The Resource Centre
• EELA Computing Element
• 5 EELA dual-core Worker Nodes
• EELA 500 GB Storage Element
• EELA User Interface
• Middleware gLite 3.1.0
• Managers J. Bogado García and F. López

• _at_ CeSPI Data Centre of UNLP
• High availability, high security, Redundant
  electrical power feed and cooling, racks and
  raised floor
• Gigabit LAN, 20 Mbps link to Internet
• 4 Mbps to Clara through a symmetric link to
  InnovaRed

10
The UNLP RC in EELA-2
Very modest resources but the first ones in
Argentina in a pro-duction Grid infrastructure.
Very modest resources but the first ones in
Argentina in a pro-duction Grid infrastructure.
11
Joint Research Units in EELA-2

• A Joint Research Unit (JRU) is a partnership for
  e-Science Collaboration, with no legal status as
  such, formed between entities of the same
  nationality and meeting the following conditions
• scientific and economic unity (internal MoU in
  place)
• lasting a certain length of time (not linked to a
  particular project)
• recognised by a public authority (minister,
  secretary of state, etc.).

No JRUs were established in LA during EELA
One of the major successes of EELA-2
12
EELA-2 achievements - NA3
Applications distribution per country, continent
and scientific domain (Y1)
2
12
13
Why is it possible?

• The EELAs
• Trained users, tutors and administrators in LA
• Disseminated the Grid culture
• Approached Decision Makers
• Proposed a model for long-term sustainability of
  Grid infrastructures in LA
• Are currently looking for funds in EU (EC
  CIEMAT) to help implementing the model during the
  next 2 years
• Conclusion
• It should work!!!!!!!!!!!!!!!!!!!!!

14
Last but not least

• Before the GRID
• I
• After the GRID
• We ..