LCB PEP
LCB Document Reference No.
CMS Addendum to CERN/LCB 97-10
CMS contribution to RD45 for the period
June 1997 December 1998
PROJECT EXECUTION PLAN
Prepared by
CMS Collaboration
Contact Person:
Vincenzo Innocente
CERN ECP/CMC
25/11/1997
CMS contribution to RD45 for the period
June 1997 December 1998
PROJECT EXECUTION PLAN
1.0 INTRODUCTION
As specified in the CMS Computing Technical
Proposal, an ODBMS plays a central role in the
CMS computing model.
It is therefore natural that most of the
research and development program on CMS software
involves studies about an ODBMS and how it can be
used to solve our problems.
CMS is currently engaged in four sub-projects
which involve the use of an ODBMS and are strongly
related to RD45 activities:
- GIOD Project;
- R&D on data retrieving;
- Prototype of the Event Data Model and of a Reconstruction
and Analysis Framework;
- Prototype of a database for Asynchronous data (calibrations).
2.0 OBJECTIVES
The broad goal of these sub-projects is to understand the impact of
the use of a ODBMS on the CMS experiment in areas
like software development, software maintenance, data modelling,
data storage, data distribution and data retrieving.
The results of these investigations will be provided
to the CMS experiment, and more in general to the HEP community,
in form of presentations to CMS meetings, RD45 workshops and
computing conferences and written reports published on the WEB.
The code and its documentation produced as the result of prototyping
activity related to these studies will form the base for further developments
of CMS software.
3.0 WORKPLAN
3.1 Assumptions
It is assumed that Objectivity, and the other software products part of
LHC++ which are used
in the sub-projects, are provided and installed by RD45 and/or IT/ASD.
4.0 DETAILED TASKS AND SUBTASK DESCRIPTIONS
4.1 GIOD Project
The "Global Interconnected Object Databases" project is
a Caltech/CERN/HP joint project with the aim of
constructing a large-scale prototype of an LHC Computing Centre.
It is fully documented (including milstones and
resources) in http://pcbunn.cithep.caltech.edu/
where its relationships with RD45 are also detailed.
We refer the reader to this document for further information.
Two persons from CMS are presently
working on the project at Caltech.
4.2 R&D on data retrieving
To obtain high performance in retrieving data it is
critical that the maximum amount of useful data can be
read in each single I/O operation.
It is also critical to have a good caching scheme
so that the most required data are closest to the user.
In CMS there is a project to study how to obtain
high performance in data retrieving from a ODBMS
focusing on
- R&D on data clustering and reclustering;
- R&D on parallelisation of I/O intensive physics jobs which use
the ODBMS;
- R&D on data management in regional centre scenarios.
Milestones for 1998 are:
- June 1998: A prototype of a reclustering algorithm integrated
with the TagDB classes form LHC++/HepODBMS
- December 1998: Prototype of strategies for data organisation and access
Two persons from the ECP/CMC group
are working on this project for a total of 1FTE at CERN.
4.3 Prototype of the Event Data Model and of a Reconstruction
and Analysis Framework
The major difference with respect to previous experiments is indeed in
the use of an ODBMS to store the event data and in using OO technique
to build the reconstruction and analysis software.
CMS is presently building a prototype of an OO Reconstruction
and Analysis Framework which uses Objectivity/DB to manage
both event and detector data.
This system is presently under development and test
using test-beam data.
Milestones for 1998 are:
- June 1998: Prototype of a reconstruction framework
for test beam and simulated data
- December 1998: Release of a "User analysis environment"
Two persons from the ECP/CMC group are working on this project
for a total of 1FTE.
We expect the person-power assigned to this project to increase
to at least 2FTE next year.
4.4 Prototype of a database for Asynchronous data (calibrations)
This project is aiming at building a database application
for detector data. It uses the clock-time to associate
detector data to event data.
The first phase of this project ended in November 1997.
Three more months in 1998 will be required to package the product,
produce user documentation and port it to new versions of LHC++ libraries.
Milestones for 1998 are:
- March 1998: Integration of the present prototype with
Objectivity Version 5
- June 1998: Release of a class library and documentation
One ECP/CMC technical student
has been fully devoted to this project for 14 months.
5.0 DELIVERABLES
The major deliverables for each sub-project
are reports of the results of the studies
in a form publishable on the WEB.
6.0 RESOURCES and SCHEDULE
Table 1. CMS Resources in Person-Month
|
|---|
| 1sem97 | 2sem97 | 1sem98 | 2sem98 |
|---|
| GIOD | 0 | 12 | 12 | 12 |
| R&D on data retrieving | 6 | 6 | 6 | 6 |
| R&A Framework | 6 | 6 | 12 | 12 |
| Asynchronous data | 6 | 4 | 2 | 1 |
7.0 REFERENCES
- RD45 PEP http://wwwinfo.cern.ch/asd/cernlib/rd45/reports/rd45_pep97.htm
- CMS contributions to latest RD45 Workshops can be found in
http://wwwinfo.cern.ch/asd/cernlib/rd45/workshop/agenda_july97.htm
http://wwwinfo.cern.ch/pl/cernlib/rd45/workshop/agenda_mar97.htm
- GIOD Project http://pcbunn.cithep.caltech.edu/