[ Home | Contents | Search | Post | Reply | Next | Previous | Up ]


Software panel of LHC computing review: Further questions (lhcb-comp)

From: Pere.Mato@cern.ch
Date: 4/14/00
Time: 1:49:19 PM
Remote Name: 137.138.142.33

Comments

LHCb got a list of further questions from the software panel of the LHC
computing review. These questions need to be answered in written by the 30th
on April. I am appending the questions in this mail. I  have attached to
each question the name of who should feel responsible of providing the first
draft answer. All comments and suggestions are welcome.
Pere
Questions to all experiments and CERN/IT
1. What are your expectations and recommendations how to optimize the use of
common software products and solutions for more than one experiment? How
should the various phases of a potential 'project' be started, resourced at
the appropriate level, how should management oversight be structured and
what kind of regular interaction would be helpful? (==>John Harvey)
2. Please comment on the idea of software agreements and MOU's, what
granularity of responsibility you would recommend, how should the resources
be shared? (==>John Harvey)
3. Please provide more detailed plans on
? project task breakdown
? resource loaded schedules and milestones
? resource profiles up to 2005, especially personnel (including an estimate
on software professionals and physicists
? contributing to software effort)
? estimate of incremental people (e.g. for 2006)
How many people contributed to software development in 1999 and 2000? How
much increase of resources are you planning for the future?  (==>John
Harvey)
4. What is criteria for accepting a new third party package into the
experiments software system? (==>Pere Mato)
Questions/Requests for all Experiments
1. Please formulate a statement expressing the experiment's requests for the
future evolution of the GEANT4 collaboration and project.  Explain how this
will meet your needs for validation, further development and ongoing
support.
(==>Florence Ranjard, Vanya Belyaev, Gonzalo Gracia)
2. What part do you expect/hope to play in the evaluation of the results of
Espresso and the decision process for what should be done about providing a
common Object Storage solution for one or more LHC experiments? (==>Pere
Mato)
3. Please provide a brief statement of what you would consider to be the
scope and deliverables for a common project aimed at a Geometry Description
system.  What would be your goals in participating in such a common project
and what roles would you be prepared to play? (==>Rado Chytracek, Pere Mato)
4. As a result of participating in this software panel, where your
representatives have heard details of work going on in all of the other LHC
experiments, have you
? identified any new areas of commonality?
? formulated any opinions on what forums would encourage continuing dialogue
and interchange of ideas and software between experiments?
? identified any areas where your experiment could/should use software
authored by one of the other experiments or IT division?
? identified any new concerns or areas of risk?
(==>Suggestions from everybody)
5. Data Management has to be as an engineered system, it involves a
component of a program framework, the bookkeeping and logging databases, the
hardware systems and networks, and the interactions of the different parts.
It involves machines/disks/robots/tapes/networks and be subject to the wider
effects of transaction management and sharing of certain central resources.
Who is designing this system and how are the responsibilities of the
different aspects of the data management and data handling distributed? How
are these people interacting with those determining the framework,
architecture, and persistency mechanisms? How are you ensuring that the
broader overall data management system will be carefully designed and will
function well? Which role do you expect from CERN/IT?
(==> Pere Mato)
6. How big is the participation of physicists from the collaboration on
software development? How are you trying to control unnecessary duplication
of software developments?  (==>John Harvey)
7. Do you consider that the "traditional architecture" can not cope with the
LHC-detector and experiment complexity? Why? What the architecture has to do
with the complexity of the collaboration? (==>Pere Mato)
8. Is your architecture flexible enough to allow the co-existence of "action
on demand" and "explicit invocation" in the same application?  (==>Pere
Mato)
9. Does your persistency solution restrict you on what classes you might use
to implement a given physics module?
(==> Markus Frank)
10. Within your architecture, are you able to migrate a physics module from
the reconstruction environment to the trigger environment (event filter)? Is
the event data presented to the event filter algorithm in the same format?
(==>Pere Mato)
11. Would you be ready to support other experiments using your tools? Would
the experiments be able to come with new requirements? Do you think it is
possible to have a common tool supported by CERN/IT?
(==>John Harvey)
Questions/Requests for LHCb
1. Has LHCb had any external review of its Architecture and Framework to
establish the cost and risks of using such a general and open approach?  Are
there any estimates of the performance costs, development time costs, and
potential debugging costs associated with the approach?
(==>Pere Mato)
2. If LHCb is really prepared to 'use what is provided' as a common solution
for data persistency (and indeed a data management system) how will you be
assured that it will meet your needs?  Are your requirements for a data
management system formulated and written down? What will be the process to
come to a list of requirements to decide on the persistency solution ? (Same
as question 2 to all experiments, ==>Pere Mato)
3. Given the architectural choices of LHCb what constraints will be put on a
decision to use Java?  Will the experiment establish some rules or guidance
for language choices, persistency choices and scripting language choices?
(==>Pere Mato, Stefan Probst)
4. Do you believe that your choice(s) for data persistency will place
constraints on your analysis tools and if so what will that mean for the
experiment?  (==>Gloria Corti)
5. How do you foresee to maintain consistency across persistent objects
stored in different technologies? (==> Markus Frank)
6.  How do you plan to support schema evolution and maintain the converters
used? (==>Markus Frank)