The main aim of the project is to create a METAComputer (or
virtual computer) that will allow utilisation of
hardware equipment,
already installed as part of the http://www.ics.muni.cz/research/Supercomputing.html">Supercomputing
project, and, at the same time, to solve problems whose requirements
(on memory, main processor, etc.) exceed
capacities of individual nodes.
Despite the different vendors, operating systems, or physical locations of
individual computers, the users
should gain a uniform access to all resources, without even the necessity of repeated login.
In parallel, it will be necessary to build an environment
allowing users to perform extensive distributed
computations utilising computer capacities from more than one location.

Verification of the usability of wide area high performance networks,
built under the TEN-34 CZ program, for distributed computing, is a complementary aim of this project.

The following partial aims can be identified:

• Physical interconnection of all participating locations by a high-performance network.
  
• Creating a unified distributed filesystem providing for
  easy and transparent work with local data on remote computers.
  At the final stage it is expected that concepts of local and remote data will lose most of their current sense.
  
• Unifying system login techniques (both on administrative and technical levels). The aim is a unified administration of the user space so that no repeated user authentication is needed.
  
• Creating a unified batch system of task administration, utilising information about the current load of computer performance.
  This task is closely related to the problem of the co-existence of CPU-expensive
  tasks, administered by batch systems, with interactive tasks requiring fast
  system response during the visualisation phase.
  
• Installing and testing (at least on a prototype level) remotely administered visualisation systems - specifically,
  the AVS system (at CTU in Prague) and IRIS Explorer (at MU-ICS). At the same time, other special visualisation systems, e.g., SPARTAN
  and InsightII (BIOSYM) in computational chemistry will be
  tested intensively. Concerning visualisation of outputs obtained by various techniques and means, a close co-operation is expected with the Masaryk University Faculty of Informatics where a Human-Computer Interactions
  Laboratory specialising in virtual reality is being established.
  
• Support of distributed computing, both in the sense of software development
  (accessibility of all compatible means through the whole METAComputer) and perfoming tasks (via batch systems) with load balancing.
  

Besides creating the METAComputer, the project has a pilot character as well:
it is expected that the experience obtained from building and running the METACentre
will allow extensions of the METAComputer system by other computers.
In a long-term perspective it is necessary to plan for connecting a high number of less efficient computers (workstations, high-end personal computers)
and continuous merging the METAComputer with institutional networks.

Results

Publications