Summary of meeting with ALICE on 13/12/02
Federico Carminati, Rene Brun, Predrag Buncic, Fons Rademakers, Peter H.
Alice is just finishing its Physics Performance Report for which
a major distributed production effort was undertaken.
To manage wide-area distributed computing Alice has develop Alien.
Alien is currently using the Globus toolkit and a large set of open-source
Perl modules for process intercommunication, database management and
It offers to Alice physicists access to Alice distributed
resources through a web portal as well as unix command-line and
C++ programmatic interfaces.
Alien has full control of the resources that it is asked to manage:
Alice physicists do not require to be involved in the details of
the resource management and job scheduling.
Short-Medium term plans
Alice plans to further develop and consolidate Alien in the near
future: Alice will keep the current Alien user interface and
architecture while it will consider, case by case, to replace the
current backends with new grid middleware products.
At start, LCG-1 will be seen by Alice just as one more site.
Alice will soon deploy Proof (parallel root) on privately managed farm.
Alice will not object (but is not pushing) in having Alien as a common LCG project
provided that the current architecture and development plan is not
Alice simulation, reconstruction and analysis software is based on the
AliRoot framework which depends directly, with no
intermediate layer, on the services provided by ROOT.
The very same applications are used in batch and interactively.
All interactive analysis goes through the "Root shell".
A strong synergy exists between Alice and Root and this makes sure
that many tools developed in Alice (the Virtual MonteCarlo for instance)
can be moved into Root with ease.
Alice is convinced that this synergy guarantees that the evolution of Root
will cover all Alice needs for the interactive environment as well as for the
In particular Alice expects to take profit of any new software through a proper
interface to Root.
Alice plans to use year 2003 to consolidate its software system to be ready
for next large data challenge early 2004.
Alice strategy is to use directly Root. No other tool is
foreseen to be used. Therefore there is no need for any common or
abstract interface to data analysis.
Alice do not recognise any role to python at physics analysis level.
Alice physicists consider to have made already a major effort in learning
C++ and do not consider worth learning one more language just to do what
they can already do with C++ inside root.
For what standard scripting application are concerned, Alice uses mainly perl
and does not see any advantage to move to python at the moment.
Alice does not currently have a fully interactive 3D event display.
Root allows to export 3D scene to OpenInventor, no further
interactivity is possible.
Alice would not object to a common 3D event display framework and
toolkit provided that it can be integrated in the Aliroot interactive
environment without interfering with Root "event-loop".
Alice considers the speed by which the grid projects are advancing by far
to slow particularly when the large resources invested are taken into account.
In contrast, they were themselves surprises by how fast Alien (one person effort)
was deployed and used by physicists fully satisfying current Alice needs.
Alice physicists are fully satisfied of their current analysis (interactive and
batch) environment and are confident that the planned development of Root and Alien
will match their future needs.
Alice may be interested in a common "event visualisation software" and to
have a LHC-wide collaboration on Alien.
Alice considers that any common project on "physics interface" will NOT cover
areas on their critical path that are not already addressed by Root or Alien.
includes pointers to AliRoot and other software projects
Most recently modified on Mon Jan 13 09:53:47 MET 2003 by Vincenzo Innocente