You can schedule jobs to run:

• Hourly, at the specified minutes past the hour. A job cannot be scheduled to run more than once an hour.
• Daily, at a specified time.
• Weekly, on one or more specified days a week.
• Monthly, on a specified day of the month.
• On a custom schedule, using a crontab expression. See Supported Crontab Expressions below. A job cannot be scheduled to run more often than every 60 minutes.

To schedule a new job:

1. Navigate to Admin Center > Schedules.
2. Select New Schedule.
3. In the Schedule Identification section, enter a display name and optional description and then select Next.
4. In the Job Details section:
   1. Select the project and job type.
   2. Depending on the job type, choose the source dataset to refresh, data product or workflow to run, or publish destination.
   3. Select Next.
5. In the Frequency Details section choose the frequency at which to run the job. If you are entering a custom value, see Supported Crontab Expressions below.
6. Select Create Schedule to schedule the job.
   This job appears in the Schedules table.

Supported Crontab Expressions

When scheduling a job to run at a custom interval, you supply the timetable for the scheduled event following the Unix-style crontab format. Valid crontab values are:

• Minute: 0-59
• Hour: 0-23
• Day: 1-31
• Month: 1-12 or JAN-DEC
• Week day: 0-7 (0 and 7 are both Sunday) or SUN-SAT
• In addition, you can match multiple values, as follows:
  • Match all values: *
  • Match a range of values: 1-5
  • Match a range of values with step: */2
  • Match a list of values: 1,2,3 (comma separated, no spaces)

Crontab Expression Examples

Every day at noon and midnight

Crontab expression: 0 0,12 * * *

Every Monday at 9:30am

Crontab expression: 30 9 * * MON

Every weekday at 9:00am and 3:00pm (15:00)

Crontab expression: 0 9,15 * * 1-5

The first day of each month at 8:30am

Crontab expression: 30 8 1 * *

Best Practices for Automating the Data Pipeline

If you are scheduling jobs to run a data pipeline (for example, refresh the source, then run the data product, and then publish the results), leave a buffer between each job.

Monitor how long jobs take to complete, and use that information to ensure you have an adequate buffer between jobs. The Schedule Events page for the scheduled job provides the average duration of the job over the past 30 days (see Managing and Viewing Scheduled Jobs).