Title: CMS Prototype Calibration DB(s)
1CMS Prototype Calibration DB(s)
- Jack Cranshaw
- for EMU/HCAL groups
- October 6, 2003
- FNAL-CD Briefing
- HCAL Prototype Database
- EMU/HCAL Plans for Slice Test 2004
2- Required functionality
- Values for calibration constants and their
precision (error) - Values for pedestal measurement and its
precision (error) - Must have support for different types, methods
and algorithms for - measurement and calculation of calibration
constants - Channel identification
- Must be able to organize data in separated sets
3- Logical structure of the HCAL database
- Data calibration constants, pedestal
measurements. - Administration management of the data sets and
types - Identification channel identification
- Systems data is divided in subsystem sets
HB, HE, HF, ...
4External Hardware Database
Systems
Administration
Data
Ch. Identification
HE
HB
HF
5ROOT Analysis
Sets management
CMSIM
ORCA (Object Oriented Reconstruction for CMS
Analysis)
Read
Write
Interface
Calibration (ASCII)
Write
DAQ
Reco Test beam 2002/2003
Read
Root Analysis
Oracle DB
C
6HCAL Planned Upgrades
- Review design.
- Include other elements of HCAL for current
calibrations. - Include full set of calibrations for jet
reconstruction. - Look at extending it to other detectors.
7Plans for Slice Test 2004
8A. Tasks Clearly Under Our Control
- Prototype HCAL calibration development.
- Should we use code generation?
- Would like advice from CD on this.
- Evaluate ConditionsDB interface for our needs.
- What do we do with outside calibrations?
(voltages, ..) - Can we handle non-Oracle files transparently?
- Could be important for future choices.
- Where do we keep raw data and in what format?
9B. Problems Needing a Working Decision
- What is the organization of the CMS database
efforts? - Prototype organization formed around Frank Glege
at 9/2003 CMS week. - Which users in the near future? Far future? When?
- Which environments at which periods?
- How much do we have to fudge because its not
ready yet? - Most of it is already fudged. Need to fit it
together. - How many database formats will be in use?
- Oracle, Text, Root, Xml, Mysql,
- When does the database become non-development?
- Too soon to buy real production hardware. Would
be happy with plan. - Can we cobble together a full dev/int/prd setup
for testing later in 2004? - Are there any legal or budget issues associated
with software or licenses? - Can CERN-FNAL negotiate that? Is it already done?
- Where do we get DBA support from?
10C. Tasks Requiring Slice Test Coordination
- Which detectors will be using the database?
- Which software will be in use which needs the DB?
- Who is responsible for the various pieces?
- Investigate middle tier software.
- Possible CD-LCG collaborative project if
generalized. - Which servers are in place?
- Development only?
- At CERN, FNAL, both?
- May need DBA help for CERN side as well
- Do we want to do any basic export/import or
replication? - Where is the master? the copies?
HCAL, EMU having regular meetings. Still
evaluating status.
11General Structure Databases
- Assume that a single development server is
sufficient. But where is it? - Will need Configuration, Calibration, Hardware
- Can we handle mixed database formats ok?
- Is there a way to translate?
- Do we consolidate?
- DAQ database access. Directly? Flat files?
- What software do we want to have in place and
test? - Do we try to distribute?
- How much support from CERN/FNAL computing?
DAQ
Mysql
slow
txt
chips
Crates
Software
xml
geom
Data
oracle
calib
int
12How We Are Proceeding
- Assume this is a valid subset, and start
assigning names and dates. Expand or re-order as
needed. - Assume at this point that we are free to fudge
anything that we need from someone else that is
not going to be available. - Assume that any software developed is extremely
useful, but is not the final product. - Go ahead and start looking past slice test.
- Need to build up DB design expertise.
13Additional Slides
14Application Program
ORCA, Reco (TB) , etc.
User's Methods
Single class HCAL_Calibration
HLM
Level 0 condition (offers opportunity for easy
creation of large variety of I/O methods)
Level 1
API Basic I/O
Level 0
Basic I/O in general form
DataBase
Oracle DB