Summary of User Requirements for Calibration and Alignment Database Magali Gruwé CERN PH/AIP ALICE Offline Week Alignment and Calibration Workshop February 22 nd 2005
2 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Layout Context Call for User Requirements Outcome of questionnaire Answers Remaining questions Conclusion
3 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Context The ALICE off-line group is: Compiling an extensive set of user requirements Evaluating the various data sources, relation with other DB or raw data Estimating the volume, update frequency, access pattern and run-time environment We need to have a fair understanding of : Use cases DB content Implementation of tools for DB population, query, information retrieval and communications with other DB Definition of development responsibilities Design and building of the necessary computing infrastructure
4 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Layout Context Call for User Requirements Outcome of questionnaire Answers Remaining questions Conclusion
5 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Call for user requirements Essential parameters of the Alignment and Calibration DBs: Access pattern and frequency Only offline reconstruction? Typical record size Very difficult to estimate from first principles Frequency of updates We expect the majority of the updates to be infrequent This probably will not be the case Validity range parameters Run, event, time, version – what else? These all have to come from the user community More details/explanations in next slides…
6 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Questions to be answered by detector community (I): Context: The alignment and calibration constants will be ROOT files, registered in a distributed file catalogue. The files will have proper tags, described in the AliEn (gLite) metadata catalogue: validity period – run, event, time Version Questions: What type and size of Root objects will be needed for the alignment and calibration procedures? For example: One calibration object could be a set of gain constants of a calorimeter, written in a root file. This object will have a certain size and will be valid for a given period of time, until the next calibration is taken or the present one is refined (new version).
7 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Questions to be answered by detector community (II): Questions (cont’d) How frequently will these root objects have to be updated (new version produced)? How frequently will these root objects have to be accessed? What kind of and how much access to external databases will be needed? External databases could be: Construction database (DCDB) ECS DAQ DCS Trigger HLT
8 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Layout Context Call for User Requirements Outcome of questionnaire Answers Remaining questions Conclusion
9 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Answers to questionnaire Answers received from all 14 detectors Some by mail Last ones during the workshop Thank you all!! But: Some are still preliminary: Need update Most are incomplete: Need to be thought about and completed Some show misunderstanding: Need clarifications from both sides
10 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Outcome of questionnaire (I) Three types of objects: Calibration related Alignment related Dead/bad channel maps Note: Some detectors specify “one object per channel” This will NOT happen: there will be ONE object for all channels Examples: One “gain” object being an array containing all gains of all channels One calibration object being and array of all gains and pedestals of all channels
11 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Outcome of questionnaire (II) Size: (rough counting, from information sent by ) ITS: ~80 MB TRD: few MB PHOS: ~1MB EMCAL: ~1MB RICH: ~1MB? CRT: few kB MUON: ~70MB ZDC: few kB V0: 2.4 kB PMD: ~9MB TOF: ~7MB TPC: ???
12 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Outcome of questionnaire (III) Update frequency: In general: anything between 1/run and 1/year Note: Some parameters vary quickly, but update frequency will not be higher than once per run. Stored objects would be histograms for example. Example: One drift velocity object being a histogram with the time dependence of the drift velocity within a run
13 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Outcome of questionnaire (IV) Access pattern/frequency: For most detectors: no answer, or there seems to be confusion with update frequency Questions: How often do you need the data? Only in reconstruction jobs? Is there an iterative procedure? Clearly, there is better understanding thanks to workshop
14 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Outcome of questionnaire (V) External Database access: Very little input from detectors: Most detectors said: No external database access needed Or did/could not answer Other detectors: FMD: Need to know how the detector was operated at the time of data acquisition, beam species, etc… ITS: Construction/geometry/calibration databases??? TRD: ESD only? CRT: Need run header ZDC: General run conditions Machine Database! V0: General trigger T0: DCDB
15 Magali Gruwé CERN PH/AIP ALICE Offline Week, February 22 nd, 2005 Conclusions Thanks again to detector representatives for the answers to the questionnaire s/discussions/presentations We received a lot more information during the workshop than via s These give us an excellent starting point Problem with respect to size of objects needed for some detectors: Need to find a way to reduce it A note is being written summarizing our understanding of detectors requirements. It will be circulated soon. Home work for detector groups: Think again about requirements (with the highlight of the other detectors’ presentations) And more specifically about External Database access