Meeting on LHCb Computing stratgey (lhcb-cb)
From: John.Harvey@cern.ch
Date: 4/21/00
Time: 5:29:54 PM
Remote Name: 137.138.142.33
Remote User: lhcb
Comments
MEMORANDUM
----------
To : Collaboration Board, Technical Board, Computing Model Working
Group
From : John Harvey
Subject : Meeting on strategy for deploying LHCb computing infrastructure
-------------------------------------------------------------------------
I should like to call a meeting for Monday May 8th to begin a discussion
on the medium and long-term planning of LHCb Computing. The meeting
will be an opportunity to discuss our computing requirements between now
and 2005 and our strategy for deploying the collaboration-wide computing
infrastructure. We need to start to formulate our plans for acquiring the
necessary resources (cpu, disk,...) and for making realistic tests of the
systems that will be used to satisfy all our data handling needs.
In the very near future the 4 LHC experiments will be requested by the
CERN management to describe their plans for Computing, including sharing
of costs and responsibilities between LHCb institutes in a Computing
Memorandum of Understanding (MoU). This meeting is being held now to
explain in more detail the thinking behind this request and to start to
prepare our response. More background information on the subject is appended
at the end of this mail after the agenda.
Anyone who is interested to attend this meeting is welcome. Moreover,
each LHCb institute, regional centre, and sub-detector group that feels
concerned by these issues is strongly encouraged to ensure that they are
represented at the meeting by the appropriate person.
Comments and suggestions are as usual most welcome.
John Harvey
Preliminary Agenda
------------------
Date and time : Monday May 8th at 14.00
Room : to be announced
1. LHCb computing requirements 2000-2005
- physics needs - production of simulated events
- computing needs - Mock Data Challenges to test infrastructure
2. Proposal for a prototype in 2003 of the collaboration-wide data
processing infrastructure
- motivation for and description of the prototype
- identify centres that will participate (e.g. CERN, RAL, CCIN2P3/Lyon,
INFN,..)
- resource planning at each centre in terms of expected evolution of cpu,
disk,
tape, network connectivity, etc.
- resources in terms of manpower and cost
- collaboration with other LHC experiments in production of, and sharing
of, prototype
(input from regional centres and institutes wishing to participate
essential here)
3. Assignment of responsibility for software
- manpower needs for software frameworks, support and detector specific
software
- participation to core software and support activities by LHCb institutes
(input from institutes and sub-detector groups essential here)
Background Information
----------------------
There will be a meeting of two of the LHC Computing Review Panels
on May 12th at which we have to make a report answering a number
of questions. These panels are now proposing that the four LHC
experiments prepare an Interim MoU for computing by October 2000.
This interim document should describe plans for producing a prototype
of the final world-wide deployed computing infrastructure that will
be used for offline data processing of simulated and real data.
It is assumed that this prototype will be ~20% of the full scale
system and that it will be assembled and used in 2003. The prototype
would be a joint venture by the 4 LHC experiments, who would share
in its development and in its exploitation. It is assumed that the
final experiment specific project will be descibed in a final MoU
for computing which will be expected in the year 2003 i.e. after
the Computing TDR has been produced.
For the meeting of May 12th we are invited to comment on this proposal
and to make a first go at describing the contents of the interim MoU i.e.
- to describe the prototype in terms of the regional facilities and
institutes relevant for LHCb. It is assumed that the scale of the prototype
should be substantial so that realistic scalability tests can be made
(~50% of the total requirements of one of the large LHC experiments) and
that the substantial cost will be amortised by the sharing of the prototype
between the 4 experiments.
- to describe the goals of the prototype i.e. to make tests of the
complete data handling chain including analysis
- to define the sharing of the responsibilities and costs amongst the
various centres and collaborations
- to define the tasks, schedule and milestones
It is expected that the Computing MoU will also attempt to assign
responsibilities for development and maintenance of software. A first
evaluation of requirements and resources has indicated that:
- ~20% of the total effort requires software engineers responsible for
the development of the software frameworks and for support (librarian etc.)
All experiments have indicated there to be a shortage of between 7 and 15
people per experiment (in our case 7 people). The MoU should address how
the collaboration intends to resolve this shortage.
- ~80% of the total effort is for physics software and it is assumed that
there
is sufficient physicist manpower to cover needs. The MoU should describe
the responsibilities taken by institutes for detector specific software
including front-end software, calibration, monitoring, event reconstruction
and simulation.
John Harvey
John.Harvey@cern.ch
tel : [41] 22 767 7358
fax : [41] 22 767 9425
|
