180 likes | 194 Views
Learn about the evolution of DATE database configuration in the ALICE Data Acquisition System, outlining the transition from ASCII files to MySQL in 2005 for improved data distribution and integrity.
E N D
Alice DAQConfiguration DB Sylvain.Chapeland@cern.ch
TDS TDS ALICE DAQ Architecture Rare/All CTP BUSY BUSY LTU LTU DDL H-RORC HLT Farm TTC TTC FEP FEP FERO FERO FERO FERO Event Fragment 163 DDL 343 DDL 10 DDL 10 D-RORC 10 HLT LDC 425 D-RORC LDC LDC LDC LDC LDC 200 Detector LDC Load Bal. Sub event Event-Building Network EDM Event 50 GDC 10 DS GDC GDC GDC DS DS GDC File PDS Storage Network 25 TDS
TDS TDS ALICE DAQ Architecture Trigger High-Level Trigger CTP LTU LTU HLT Farm TTC TTC FEP FEP Detectors FERO FERO FERO FERO LDC LDC LDC LDC LDC Event-Building Network EDM GDC GDC GDC DS DS GDC PDS Storage Network Data Acquisition
DATE v4 - configuration 2004 • DATE Configuration stored in ASCII files • $DATE_SITE_CONFIG • Shared by NFS • DATE DB package used to read static information: • roles (who is what) … • detectors (what detectors and sub-detectors are made of) … • triggers (what is activated by each trigger) … • event building (what to do with the events on GDCs) … • memory banks (where to store the data structures) • Readout equipment (data sources on LDCs) • RunControl parameters (dynamic) • Misc. items: DIM DNS, environment, …
DATE v5 - configuration 2005 • SQL based • Constraints to avoid inconsistencies • GUI • APIs unchanged • Backward compatible (runtime/compile switch) • Critical shared resources distributed with SQL • Few static items still in flat files 2007: Now everything in MySQL
Operator console MySQL server DATE nodes Architecture
SQL db • Improves data distribution and configuration integrity compared to NFS ASCII files. • Simple structure, few tables (10s), little data (1000s records) • Peak read access when DAQ starts (100s clients) • Infrequent write access (by operator and runcontrol) • No use of special features, standard SQL, any back-end should fit • DATE Language interfaces: C, Tcl/Tk, php
SQL db - requirements for development phase • Ease of installation • Dynamic setup, may need to create many times different systems • Independency for quick changes • Control on the server: • to monitor behavior • how to tune performance • improve server setup accordingly
SQL db - requirements for production phase • Version control • Version must be stable during data taking period • Maintenance and operation • Fast reactivity, known procedures, internal know-how • On the fly setup change when needed (performance tuning, …)
MySQL 2005 • Open source • Free DB system to distribute to institutes using DATE • Easy setup • Good feedback on bugs (#7838) • Platform compatibility • Interfaces for C, Tcl, Php … • Good performance • No strong dependency, in house expertise • MySQL version 4: innodb, … • MySQL version 5: promising features (views, triggers, …)
MySQL • RPM setup, binary tarball, source compilation • Server setup: • Users and table creation with simple SQL commands • Can also copy an existing database: copy whole directory and it works (good for backup as well) • Client program: linked with corresponding platform library Installation is a matter of minutes
Operator console MySQL server DAQ nodes DB performance tests Configuration DB: concurrent access 65 LDC, 65 GDC
DB performance tests 2007
MySQL operation • Excellent stability – no crash so far • Simplicity to setup & run • Scalable • Nice GUI tools: • MySQL Administrator • MySQL Query Browser • MySQL Workbench
Other DB usage in Alice DAQ • infoLogger store logs of all DAQ nodes • Detector Config specific hardware configuration • File Exchange Server export data to outside world • Logbook Summary information for each run • Machine db DAQ hardware configuration: LAN, cables, racks, …
Conclusion • Now entering production • Software ready and stable • Still flexible for new features • Good choice so far