Athena/POOL integration and conditions data
D. Adams
30apr03 0920
Motivation
The test beam efforts are acquiring data and looking to the database
group for guidance on how and where to store this data. One plausible
long term plan is to store it in POOL and use the IOV service to
associate times amd event numbers with POOL tokens.
Plan
Joe Rothberg, Steve Goldfarb and I sat down a sketched out a plan
to demonstrate such a system.
Joe described four kinds of test beam data:
- 1. Raw temperatures
- 2. Raw alignment scans
- 3. Calculated r-t calibrations for each plane
- 4. Calculated alignment constants (position and orientation) for
each chamber
Joe envisions that the programs that collect the raw data and those that
perform the calculations are not in C++ and would reside on different
machines than the one responsible for storing the data. I call the collection
and calculation programs producers and the storage program the recorder.
The data that these programs exchange are C-structs.
Design
The following components will be created and integrated to move the
data into POOL and register it in IOV.
- 1. A C-struct representation of each type of data.
- 2. A network communications system to transfer C-structs using a
transmitter and a receiver.
- 3. A producer for each type of data including the above transmitter.
- 4. A C++ representation of each data type suitable for storage in pool
and for accessing data in athena.
- 5. A collector that does the following
- receives data as a C-struct
- constructs the corresponding C++ object
- stores the object in POOL receiving a token
- registers the time and token in the IOV system
When the above system is in place, it should then be possible to
use the existing Athena/IOV integration and the upcoming Athena/POOL
integration to read the data back into Athena. Or to retrieve the data
either by token or time in other applications.
People and schedule
Joe agreed to take care of items 1-3 in the above list. Steve will provide
item 4 and then David and Steve together will provide item 5. We did not
yet agree on a schedule but a gaol of storing at least one type of data
by some time in April seems reasonable.
Other systems
Other than the actual data, there is nothing in this plan that is specific
to the muon system and I can imagine doing something similar for other
subdetectors either in parallel or after the muon implementation.
Will the data be accessible in one, five or twenty years?
POOL is not promising that data stored this month or next will
accessible in future releases. However, by the summer, I expect that
POOL will provide a data format that will be accessible by all future
versions of POOL. Even if the format becomes unreadable, I beleive we can
reasonably expect means to transfer data to a readable format. It is
important that we express this requirement to POOL well in advance of the 1.0
release scheduled for July 1, 2003.
The lifetime of POOL is a separate question. It is an integral part of
the LCG project which is an integral part of the CERN plan to support
LHC computing. Both ATLAS and CMS plan to start writing event data into
POOL this year.
dladams@bnl.gov