Understand analysis scheduling

Each analysis requires you to specify scheduling. Scheduling indicates when to evaluate an analysis automatically. There are two types of scheduling:

• Periodic

  With periodic scheduling, evaluation occurs based on clock time (DateTime object expressed as Coordinated Universal Time (UTC)). You can specify a specific time for evaluation each day or the time between evaluations. With PI Server 2015 R2 and later versions, the maximum frequency you can set for analyses is 120 times per second.

  Note: If you have analyses that are previously configured to run more than 120 times per second, you must reduce the period to less than the allowed maximum, else they will generate errors while running with PI Server 2015 R2. Similarly, if you are setting periodic scheduling outside of the PI System Explorer user interface, make sure that the period is set to less than the allowed maximum.

  If the period is not divisible by the 24, 7 minutes for example, the evaluation time will reset at the beginning of each day. Let's say you have scheduled the analysis to run every 7 minutes and the last evaluation was at 11:56pm. The next evaluation will be at midnight (4 minutes later), not at 12:03am the next day.

• Event-Triggered

  With event-triggered scheduling, evaluation occurs based on events. You can specify one or more input attributes that trigger an evaluation whenever the attribute value changes.

  Note: When auto-backfilling is enabled, triggering events with time stamps before analysis service start-time are ignored for the purpose of real-time evaluation. For more discussion on real-time evaluation and auto-backfilling, including the setting of the AutoBackfillingEnabled configuration parameter, see Understand PI Analysis Service settings.

By default, analyses use event-triggered scheduling, triggered on changes to any input.

Many factors affect the speed that an analysis runs and writes output values:

• System architecture, including inputs and outputs

• Network configuration

• Load and performance of different computers

Monitor and test your analyses, especially those scheduled at high frequencies, to ensure the system resources support the configuration.