DEISA perspectives Workshop on Biomedical informatics, Brussels, March 1819 2004

1
DEISA perspectivesWorkshop on Biomedical
informatics, Brussels, March 18-19 2004

• Victor Alessandrini
• www.deisa.org

2
Distributed European Infrastructure for
Supercomputing Applications - Consortium

• IDRIS CNRS, France (coordinator)
• FZJ Juelich, Germany
• RZG Garching, Max Planck Society, Germany
• CINECA, Italy
• EPCC (HPCx project), Edimburgh, UK
• CSC, Helsinki, Finland
• SARA, Amsterdam, The Netherlands
• ECMWF (european organization), Reading, UK

3
DEISA mission statement

• To contribute to a significant enhancement of
  capacities and capabilities of HPC in Europe, by
  the integration of leading national
  supercomputing infrastructures.
• To deploy and operate a distributed
  multi-terascale European computing platform,
  based on a very strong coupling of existing
  national supercomputers. DEISA plans to operate
  as a virtual European supercomputing centre.
• To contribute to the deployment of an extended,
  heterogeneous Grid computing environment for HPC
  in Europe, needed to interface the DEISA research
  infrastructures with the rest of the European IT
  infrastructures.

4
DEISA management (operation as a virtual
Europeansupercomputing centre)
5
Artists view of the facility
The DEISA super-cluster (the core project)
Dedicated network interconnect (reserved
bandwidth)
National computing facility
Site B
Site A
Site C
Site D
Extended Grid services
6
DEISA and Grid technologies

• Large scale integration of IT systems is, for
  DEISA, primarily an strategic issue
• DEISA is technology neutral. Technology choices
  follow from their capability to adapt to a
  pre-established operational model, and to provide
  real services to end users.
• Three criteria for technology choices
• The necessity of disposing as soon as possible of
  a stable and reliable European production
  platform (the core platform). Grid technologies
  work in the background through global file
  systems and multi-cluster batch managers.
• The necessity interfacing this platform with
  other systems in Europe (the outer Grid
  environment). Deployment of traditional Grid
  environments, like the Globus Toolkit or Unicore.
• The preparation of their future evolution by the
  integration of new Grid technologies as they
  reach maturity (a specific JRA activity).

7
The first generation core infrastructure(operat
ional early 2005)
125 cabinets like the ones shown above (IBM 690,
690, 655) 4000 processors (5 8 Gigaflops per
processor) About 10 Terabytes memory A lot of
useful teraflops. Inclusion of huge Linux and
vector systems planned for 2005.
8
Global File Systems
Sophisticated software environment, necessary to
provide single system image if a clustered
computing platform.
Global file system
They provide global data management. Data in the
GFS is symmetric with respect to all computing
nodes.
GFS encapsulate sophisticated distributed
computing and Grid technologies. Applications do
not need to be modified to benefit from GFS
services. Grid technologies are working in the
background, and they are not directly seen by end
users.
9
The DEISA integration concept (core
infrastructure)
Site A
Site B
Global distributed GPFS file system with
continental scope. Global resource pool is
dynamic nodes can enter and leave the pool
without disrupting the national services..
Network interconnect (reserved bandwidth)
Site C
Site D
10
The DEISA operational model

• Global management of an European resource
  pool, that will provide
• An integrated supercomputing environment for
  trans-national collaborations, providing global
  data management through global file systems.
• No need to grid enable huge applications (unless
  they are grid enabled by design). Applications
  can benefit from important computational
  resources being run as such.
• Implementation of job migration across sites
  (transparent to end users) as a way of releasing
  significant resources in one particular site for
  demanding DEISA applications. We are load
  balancing computational throughput at a European
  scale.
• Support of distributed applications (grid enabled
  by design).
• With this operational model, the European
  super-cluster is not very different from a huge
  local supercomputing cluster (which will be
  partitioned anyhow because of fault tolerance).

11
Scientific perspectives

• We (the DEISA partners) have decades of
  experience in the production of leading edge
  computational science.
• We dispose of competent staff for user support,
  system integration,
• Integration of human competences is as important
  as the integration of computing resources.
• We have developed, over the years, very efficient
  interfaces with the different scientific
  communities (climat modeling, quantum chemistry,
  material sciences, ). Each one has its own
  specific requirements.
• We are eager to develop special interfaces with
  the Bio-Medical community, in those areas in
  which we can provide added value to their
  research.
• DEISA includes a Life Science Joint Research
  Activity.

12
Prospective services (first guess)

• Using the global distributed file systems of the
  core facility to deploy shared data
  repositories.
• Deploying distributed heterogeneous file systems
  with external organizations to dispose of
  enhanced privacy access to dada.
• Fine tunning new generation BMI applications to
  clusters of large, high physical memory, SMP
  clusters. Our intelligent operating systems can
  use large memory address space to create memory
  mapped file systems and dramatically reduce
  input/output overhead in data driven
  applications.
• Running very demanding applications in what
  concerns processing power and memory size.
• Running very demanding applications in what
  concerns execution times (hundreds of hours).
  This requires process migration and
  checkpoint-restart facilities for QoS (which are
  present in our platforms).
• Interfacing with other organizations, which may
  act as portals to the DEISA environment.
• Deploying high level Web-like services that will
  allow computational monitoring and steering of
  complex applications, hiding the DEISA
  environment from end users.

13
Portals, Web-like services
The DEISA super-cluster (the core project)
ASP services. Users connect to an
application, not to the computing
environment. Provides computational monitoring
and steering services. Technology already
available for the DEISA users.
Site B
Site A
Users are not IDRIS-DEISA aware Demanding
applications are rerouted to IDRIS
Site C
IDRIS
InfoBioGen
14
ASP services three tier architecture (for
computationalmonitoring and steering).
Firewall
Internet
Intranet
Gateway bridge between the Internet clients and
the legacy environment. Handles all AA services.
Clients connect to J2EE application running on
the gateway.
Back end server runs high performance legacy
application
15
Extended, heterogeneous data management
Global distributed file system with - Single
sign on - Intrinsic security model - Access
control - Internal cryptographic mechanisms -
Operating across firewalls - Web and NFS access
to data. (candidate AVAKI DataGrid)
Site B
Site A
Other organizations disposing of confidential
data sets.
Site D
Site C
16
Conclusion

• DEISA is focused on European computational
  science.
• Strong integration of national supercomputing
  infrastructures should be pursued, as the only
  way forward (today) for matching the US and
  Japanese efforts in this area.
• The DEISA consortium is committed to enhance the
  production capabilities of HPC in Europe, and to
  search for a very large user consensus and for a
  high impact in everyday creation of scientific
  results. This is its only measure of success.
• We are eager to work with the BMI community to
  identify the areas in which we can provide
  leading edge services adapted to their needs.