LHCb Computing Project Planning (lhcb-cb)
From: John.Harvey@cern.ch
Date: 6/23/00
Time: 3:20:35 PM
Remote Name: 137.138.142.33
Remote User: lhcb
Comments
M E M O R A N D U M Geneva, 23rd June 2000
To: LHCb Collaboration Board Members
CC: LHCb Computing Model WG members
From: John Harvey
Subject: LHCb Computing Project Planning
__________________________________________________________________________
The purpose of this memo is to keep the CB informed
of the status of the on-going planning exercise for
understanding and satisfying the long-term computing
needs in LHCb. The motivation for this planning exercise
was introduced by Ioana Videau at the CB meeting of May 25th.
It has been prompted by the on-going Review of LHC Computing,
which was setup by CERN's management in order to get
estimates of computing resources (cost and manpower) needed
to satisfy our data processing requirements. In addition
CERN management would like the sharing of responsibilities
for both software and computing to be specified in a more
formal way than has been the case in the past. Therefore
another purpose of the review has been to give an opportunity
to the experiments to give input on how this may be best
achieved. It is clear that final decisions on how all
this information will be defined in formal agreements
(i.e. the Computing MoU) will be made only after discussions
between CERN, RRB representatives and the managements of
the LHC experiments later this year.
I have now received a detailed request for information
from the Review Panel chairman (see below).
A number of Panel meetings have been scheduled through
the summer to discuss these issues, and the final meeting
is scheduled for September 19th. Several of the questions
concern policy issues and in particular are related to the
plans for computing in the regional centres.
Given the short timescale I propose to work on these issues
through the summer with the LHCb Computing Model Working Group
(CMWG). This group was setup recently by Frank Harris and
comprises members nominated for each country represented in
LHCb (see below). Each CMWG member will be encouraged to consult
with CB members from their country as well as other national
representatives with responsibilities for computing. A first
meeting of the CMWG is being scheduled during the next Software
Week on Wednesday July 5th.
A full report on the outcome of the Panel's work will be given
at the collaboration meeting in Milano in September. I hope
this approach meets with your approval, but in the meantime I
would be happy to receive any comments and suggestions you may
have.
John Harvey
------------------------------------------------------------
Attachment 1) List of current members of the CMWG :
Chair F.Harris (Oxford)
Brazil P Colrain (Rio)
CERN J Harvey (CERN)
France A Tsaregorodtsev(Marseille)
Germany M Schmelling(MPI Heidelberg)
Italy D Galli, U Marconi(Bologna)
Holland M Merk(NIKHEF)
Poland M Witek(Cracow)
Russia I Belyaev(ITEP)
Spain B.Adeva(Santiago), deputised by G.Gracia (CERN)
Switzerland P Bartalini(Lausanne)
UK A Halley(Glasgow),T Bowcock(Liverpool)
--------------------------------------------------------------
Attachment 2) Request for information from the Resources Review
Panel of the LHC Computing Review (N.B. the notes of explanation
are mine) :
1) Cost and manpower estimates to build and run the
LHCb computer facilities in 2005:
- at CERN (Tier 0)
- at Regional Centres (Tier 1's)
The problem of staffing these facilities should be addressed.
2) A cost and manpower estimate to build and run the
Tier 0 and Tier 1 prototypes expected in 2003.
3) A statement of the policy of each collaboration for the
access to the computing system (Tier 0 and Tier 1's)
within the collaboration itself.
(Note of explanation : We are expected to explain our policy for
permitting use of our Tier 1 centres by any institute in the
collaboration, especially if this involves remote access from a
different country. In particular we should explain how countries
without a designated regional facility (i.e. Tier 1) will access
LHCb data and perform their computing tasks i.e. analysis etc.)
4) Guidelines for writing the Computing MOU
(Note of explanation: responsibilities for core software development
and for staffing routine software activities (librarian etc.) need
to be defined such that the maintenance of the software system can
be guaranteed over the lifetime of the experiment. The sharing of
costs for computing resources needs to be defined with the 1/3 (CERN),
2/3 (collaboration) rule as a guideline.)
5) A draft of the MOU for building a large prototype Tier 0 facility
in 2003 to be shared by all 4 LHC experiments.
(Note of explanation : There has been a proposal, originating from
CERN/IT, to construct a large scale prototype of a Tier 0 facility
for making scalability tests in 2003. It is expected that the four
experiments will participate in the development and exploitation of
this prototype. Regional centres will also participate in tests of the
distributed computing model which will also make use of the Tier 0
prototype. No firm conclusions have been reached yet on this proposal,
but if it is to go ahead agreements on how it will be constructed need
to be envisaged.)
6) A technical explanation (justification) of the cpu and storage
requirements
7) The detailed technical guidelines to design the prototype configuration,
namely:
o complexity (number of boxes,...)
o computing power
o data access (hierarchy of data access, data locality ,...)
o heterogeneity across the different Tiers centres
o management issues (within/outside a GRID framework)
The calendar for the summer activities of Panel 3 is the following:
3rd of July
17th of July (to be confirmed)
22nd of August (to be confirmed)
19th of September (to be confirmed)
All meeting at CERN (16:00 - 20:00).
Videonconference facility foreseen.
John Harvey
John.Harvey@cern.ch
tel : [41] 22 767 7358
fax : [41] 22 767 9425
|
