The Michigan Medicine Tableau Server has a finite capacity to process background tasks like extracts and subscriptions. Additionally, we have found that some extracts are scheduled to refresh more frequently than needed based on how often the content is accessed.

To best utilize the Tableau Server capacity, we have implemented a process to automatically reschedule extracts and notify the owner when certain access thresholds are not met.

Types of Schedules

Below are the current schedule definitions:

Thresholds

We are using the following thresholds to determine if a workbook or published data source has a scheduled extract refresh that needs to be rescheduled to one that matches actual usage.

New Schedules

Extracts will be moved to the schedule that matches usage. For example, an hourly extract that was last accessed 25 days ago would be moved to a weekly schedule since it has been accessed in the last 2 months.

Important: If your extract was automatically rescheduled, you can change it. Because there are many options for some types of schedules we encourage you to pick the best option based on the data source and business need.

FAQs

I received an email saying that my workbook or published data source has been updated. Now what?

You do not have to do anything. However, you might want to:

• Notify users that the data is not updating as frequently, especially if there are subscribers. 

• Update any notes in the dashboards to reflect the change in extract frequency.

• Change the schedule to one that better reflects the data source update frequency.
  (e.g., If your database updates on Wednesday, schedule the update for Wednesday night instead of Monday Morning.)

• If the content is no longer needed, consider archiving and deleting it.

Why are we doing this?

To ensure customers have the most current data in the dashboards they use. The Tableau Server has five backgrounder processes that can each do one task at a time (extract refresh, housekeeping, send email subscriptions, etc). Prior to making this change extracts were often delayed by hours, and emails that were supposed to be sent at 6:00 a.m. were going out after 11:00 a.m.

Can I change the schedule after it has been adjusted?

Yes. We recommend you do so if your data source schedule does not line up with the one we assigned. You can even change it back to the original schedule; however, if the thresholds are not met the schedule will be updated again. 

Can the schedule be automatically changed again?

Yes. If use continues to taper off it may be moved from Daily to Weekly to Monthly, and eventually to Disabled.

What about email subscriptions?

A subscription does not count as access for this process. Dashboards with subscriptions tend to be used more frequently as the subscriptions drive access. 

Can you provide an example?

One example is an extract scheduled to run hourly but had not been accessed in about 18 months. 

What extracts does this apply to?

Any scheduled extract on the Michigan Medicine Tableau Server. 

How often are the extracts checked?

They are checked every Thursday afternoon. This is to ensure there has been no traffic for the week, but also to ensure the change can be adjusted before the weekend. 

How can I avoid having my extract rescheduled?

Make sure your dashboard is being used. If it is still in development and not rolled out to users you can move it.

Will my extract ever be changed to a more frequent schedule?

No. We will only reduce the frequency to conserve server capacity.

Can't we just get more backgrounders?

In 2019 we increased the number from four to five. It did help, but a backgrounder can consume a large amount of resources. We would love to expand the capacity of the Michigan Medicine Tableau Server, but we currently don't have funding for it.

What if the dashboard owner no longer works here?

If the owner has left Michigan Medicine they are no longer a licensed user. In this case, the Tableau Server admins will receive a notification and determine who should be notified. 

What if there are multiple schedules for one extract?

If there are multiple schedules they will be evaluated and rescheduled individually.