Upload
keira
View
53
Download
0
Embed Size (px)
DESCRIPTION
Role of e-infrastructure in supporting European Research . Kimmo Koski March 11 th , 2011 Oslo. Themes. European Landscape in ICT support for research Nordic position Central or distributed support model How to prepare for 2015 and beyond. Landscape for supporting research in ICT. - PowerPoint PPT Presentation
Citation preview
CSC – Tieteen tietotekniikan keskus Oy CSC – IT Center for Science Ltd.
Role of e-infrastructure in supporting European
Research
Kimmo KoskiMarch 11th, 2011
Oslo
Themes
• European Landscape in ICT support for research
• Nordic position• Central or distributed support model• How to prepare for 2015 and beyond
CSC – Tieteen tietotekniikan keskus Oy CSC – IT Center for Science Ltd.
Landscape for supporting research in ICT
Some trends
CSC – Tieteen tietotekniikan keskus Oy CSC – IT Center for Science Ltd.
The Complex and Confusing European e-Infrastructure Landscape
YOU NAME IT…
Synergy in services
• Benefits obtained case by case• No need to solve all at the same shot, addressing a
group of communities can be very valuable, too
Picture: originally from EC, modified by Kimmo Koski
CASE CSC: EU projects’ portfolio (total volume in PMs)
Prace 1IP15%
Deisa23%
HPC Europa 2
5%
EGI-Inspire25%
ODE4%
APARSEN6%
ELIXIR1%
Clarin1%
MMM@HPC7%
ICE2SEA1% GN3
6%
e-Infranet3%
e-IRGSP33%
EC projects portfolio 01.2011
PMs = 568
HPC
Project area
grid
data
bio medical
linguistics
applications
network
policy work
CASE CSC: EU projects’ portfolio (total volume in PMs)
Prace 1IP8%
Prace 2IP17%
HPC Europa 23%
EGI-Inspire 14%EUDAT
20%ODE2%
APARSEN3%
e-BioMedSci7%
CLARICLE3%
ENVRI10%
MMM@HPC4%
ICE2SEA1%
GN33%
Me-W3%
e-Infranet1%
e-IRGSP32%
EC projects portfolio maximum scenario 01.2012NOTE! All projects will not be funded
PMs = 980
HPC
Project area
grid
data
bio medical
linguistics
applications
network
policy work
environmental
From infrastructure to research driven projects
Network: GN3
HPC: PRACE, DEISA, EGI
data: EUDAT, ODE, APARSEN
middleware: NDGF, EGI
Application development: CRESTA
Training, user interface: HPC-Europa2
Policy w
ork: e-IRG
, e-Infranet
Community: Elixir, Claricle, ENVRI, BioMedSci
Ice2sea MMM@HPC
Science specific problem
Science specific problem
General e-Infrastructures
Emphasis to build trust
TOP PRIORITY ACTION: Moving from horizontal projects to support vertical activities and research driven projects, but maintaining the presence and role in the relevant e-Infrastructure layers
Critical times for European e-infrastructure• Will PRACE funding be sustainable after the first round
(business model)?• Will financial national contribution in EGI allow self
sustained organization in a long run?• Will there be EU-project flagships for emerging areas,
such as data, software development, education and training, green ICT and cloud services?
• Can we build trust between researchers and service providers such as national centers?
• How can we efficiently collaborate in providing e-infrastructure and related services in Europe?
CSC – Tieteen tietotekniikan keskus Oy CSC – IT Center for Science Ltd.
Nordic position
Different support systems in different countries• Finland centralized, critical mass• Sweden and Norway distributed, local matching funding• Denmark coordinates through research groups, direct
researcher involvement in e-infrastructure decisions• Common nominators: NDGF, joint work in EGI and e-IRG,
commonalities in ESFRI-participation, 3/5 included in PRACE etc.
• Need to respect national decisions and note that history has an impact to the system– Joint services with distributed resourcing possible– Requirement to fit together with the national system
Nordic opportunities
• Green ICT and datacenters, cloud computing
• Excellent education system• Resource profiling – do we all want to
repeat the same services?• Collaboration in Research Infrastructures
(both ESFRI and existing RIs)
Role in HPC• Worth thinking about:
– How many Nordic projects have got resources from PRACE?– How many even applied?
• What is the reason?– No need?– Too good local resourcing?– Software does not scale?– Something else?
• Can we risk not getting the high-end resources timely due to time consuming peer review and uncertainties to succeed
• Major training and scalable software development challenge -> should be done together
CSC – Tieteen tietotekniikan keskus Oy CSC – IT Center for Science Ltd.
Datacenter example
Finnish competitive edge – as a country(Fits to Norway or Nordic, too…)
Modern and reliable infrastructure (national power grid, roads, airline connections, data networks)
World class education system and competences on ICT & energy
Steady economical and political conditions
Cheap energy (www.energy.eu) and strongly increasing CO2 –free capacity
Cool climate and water resources
No major earthquakes (4.1 biggest ever)
No major storms or other dangerous natural phenomena
European Gateway to Russia
Kajaani paper mills & hydro & bio energy
1919
CASE Finland: next generation supercomputing• Major investments
– 25 MEUR funding granted for investment in HPC, medium range and data management
– 30 MEUR investment in Datacenter (Kajaani)– Operation costs on top of that (in CSC budget)
• Installations in phases during 2012-2014• Procurement started, decisions expected
autumn 2011
CSC – Tieteen tietotekniikan keskus Oy CSC – IT Center for Science Ltd.
Central vs. Distributed
Characteristic for Finnish model Services centralized in the national center (CSC)
• Productivity• Quality• Cost efficiency
Strong support organization • Technical and scientific support• Multidisciplinary• Diverse fields (computing, connections, contents)
Short way to decision making• Limited Company (Ltd.)• Short path to the Ministry of Education
Functional distribution of work• Cooperation between CSC and institutions of higher learning
It is not that simple…
• Both central and distributed models have their strengths and challenges– Matching funding, areal support, distance to
customers, overhead in repeating services, competence development in universities etc.
• History has an impact• Ability to agree how to divide work is a key
issue
Distributed model • Overhead in providing services, but impact in
involving local users and training local people • Slower decisions, but high commitment after
that• Overlaps in investments, but possibility to add
local co-funding• Requires typically lot of committees and
strategy papers, but that can also be useful sometimes .
Infrastructure is evolving: can weafford to renew all every four years?
Cray XT2008
Users adapt quickly: Example from CSC supercomputer history
CSC – Tieteen tietotekniikan keskus Oy CSC – IT Center for Science Ltd.
How to prepare for future, 2015 and beyond
Some proposals• Focus on balanced services, e-infrastructure at large, including
training and education • Invest in more efficient management and utilization of data• Find successful ways to build trust in ICT between research and
service providers– What to do: research– How to implement it: service provider
• Explore work distribution in Nordic level– Selected projects, well-defined targets, clear benefits
• More efficient utilization of European resources– Active Nordic participation– Joint presence
CSC – Tieteen tietotekniikan keskus Oy CSC – IT Center for Science Ltd.
Spring is almost here!
You can already see deer walking in the fields…