Software Architecture

---

The Skim1 software is organized in 5 main processes:

	INPUT PROCESS

• PURPOSE: spooling data from magnetic tape to disk
• START: manually by the user after tapes loading

ANALYSIS PROCESS

• PURPOSE: submitting input data file to the several computing nodes of the University of Colorado
• START: manually by the user

GROUPING PROCESS

• PURPOSE: collect output data files from the farm computing nodes, check them against the database, and group them into output streams
• START: automatic crontab job every 15 minutes

	OUTPUT PROCESS

• PURPOSE: spool output data from disk to magnetic tape
• START: automatic every time an output stream is filled (approximately 4 Gb), or optionally manually by the user

EXPORT PROCESS

• PURPOSE: ftp output data via internet to Fermilab
• START: automatic crontab job every 5 minutes

---