Conditions Database Andrea Valassi Pere Mato 1 June
Conditions Database Andrea Valassi Pere Mato 1 June 28, 2000 Programme-Of-Work Meeting
Overview of components Algorithms Det Elem Cond. DB Svc • Framework design issues: Det. Pers Svc Framework Services Align Transient detector store Andrea Valassi CERN – The Algorithms need not talk directly to the Cond. DB. – Within the Gaudi framework, Cond. DBSvc is the service responsible for updating the Transient Detector Store with data from the Cond. DB. • Software responsibilities: Slow Calib 2 Det. Data Svc Conditions database – LHCb-related issues are: Cond. DBSvc design and interface to Cond. DB; data type specification. – Cond. DB is being developed by CERN IT division. June 28, 2000 Programme-Of-Work Meeting
Open issues in the design • HOW to store/retrieve condition data: – framework design of Cond. DBSvc (independent of data type) • synchronisation of Transient Detector Store used by the Algorithms – detector data identification • truly LHCb-specific: separation of Align/Slow/Calib or Vdet/Ecal/Hcal/Rich. . . • general to Cond. DB-design: versioning over multiple validity ranges – interface and needed functionalities/features for the Cond. DB • general to Cond. DB-design, involves feedback to IT • WHAT to store as condition data: – may be different for Alignment, Calibration, Slow. Control. . . – may vary from Sub. Detector to Sub. Detector ® uniform implementation: store condition data as XML strings? ® also recommend a uniform description of similar data for all SD ’s 3 Andrea Valassi CERN June 28, 2000 Programme-Of-Work Meeting
Interface to IT’s Cond. DB • Good collaboration and feedback with IT DB group – Stefano Paoli, Dirk Duellmann • Parallel requirement documents: – from LHCb (Pere, february 2000) – from IT (Stefano, june 2000) • Open design issues on functionalities and interface – versioning schemas over multiple time validity ranges • tag the head vs. tag any combination of data blocks; splitting of data blocks • browsing with global tags • add new data blocks vs. update existing data blocks – internal vs. external database management • transaction management, concurrency issues – support of multiple storable data types – separation of data and meta-data 4 Andrea Valassi CERN June 28, 2000 Programme-Of-Work Meeting
Data identification and versioning 5 Andrea Valassi CERN June 28, 2000 Programme-Of-Work Meeting
Plans and timescales • Work plan for Cond. DBSvc – use cases and requirements from Sub. Detectors • concentrate at first on one data type: Alignment • identify example of SD(s) with condition data to store and algorithms to use it: ? – design and integrate software in the framework, test on the example(s) – iterate feedback to IT • Plans for Cond. DB from IT – current effort on collecting our requirements – first reasonable prototype could be ready in a couple of months from now • My personal plans. . . – 50% on LHCb from mid-August (after coming back from Osaka. . . ) – getting started with Gaudi at first (Det. Data. Svc, Det. Pers. Svc, XML. . . ) • Summing up: prototyping of Cond. DBSvc around October to December – or is the Cond. DBSvc needed for a critical deadline before? 6 Andrea Valassi CERN June 28, 2000 Programme-Of-Work Meeting
- Slides: 6