Measure an Integration's Performance

Expect to measure an integration's performance often, including while designing the integration, before promoting the integration to production, and periodically after the integration is in production.

Perform any or all of the following tasks to measure an integration's performance. If you're troubleshooting, these steps help you determine the activities that are taking the most time and analyze how many integration instances have run for longer than expected. This information can help you determine your next steps.

Determine an Integration's Typical Processing Times

You should know your integrations' typical running times so that you can more easily identify the integration instances that took longer than expected to run.

You have two options for checking processing times: review numerical times on the Instances page, or review graphical information on the Integrations page.

How Oracle Integration Calculates Duration

Oracle Integration calculates an integration instance's duration in the following ways:

  • Schedule integrations: Time that passes from start to finish.
  • Failed integrations: Time for all activities, including sending requests and rerunning the integration.
  • All other integrations: Time that passes between the triggering event and the flow finishing.

Option 1: Review Graphical Information

  1. In the navigation pane, click Observability, then Integrations.
  2. Hover over the row of an integration, and select View processing time statistics View processing time statistics.

    The Processing time panel opens. The panel shows the processing time for all integration instances for the integration.

  3. If needed, update the time filters at the top of the panel.
    The time filters appear at the top of the panel and to the right and include a drop-down filter and the ability to show data by the hour or by the day.

  4. Above the chart, select the following values to hide the lines for the values:

    • Minimum
    • Maximum
    • Standard deviation

    Only the Average value remains.

  5. Review the average processing times for the integration over time.

Option 2: Review Numerical Times

  1. In the navigation pane, click Observability, then Instances.
  2. Click Filter Filter icon, and update the filtering requirements for the page.

    For example, from the Sort by drop-down, select Duration. You might also want to update the time window filter.

  3. For each integration instance, find the Duration column, and review the value in the Duration field.

Determine an Integration's Performance on a Given Day

You can quickly measure an integration's performance for a given day.

  1. In the navigation pane, click Observability, then Integrations.
  2. Hover over the row of an integration, and select View processing time statistics View processing time statistics.

    The Processing time panel opens. The panel shows the processing time for all integration instances for the integration.

  3. If needed, scroll down until you see the Instances chart, which is the second graphical chart on the page.
  4. Point to the vertical line for a given date, such as a date with a spike in its processing times, and click the line.

    The Instances page appears and shows only the integration instances that ran on the selected date. The integration instances are sorted by instance duration, with the longest durations appearing first.

  5. Optional: Identity the activity that required a longer-than-usual processing time.
    1. For an instance with a longer-than-expected duration, hover over the row, and select View details View icon.

      The activity stream opens.

    2. Review the processing time for each part of the integration flow to determine which steps took the most time.

      The processing time appears after the timestamp, above the information about the activity.
      A gray box appears around the 97ms time, which appears on a milestone in the activity stream

    3. Identity the activity that required a longer than usual processing time.

Determine Whether an Integration's Performance Is Consistent Over Time

You typically verify that an integration meets your organization's performance requirements before promoting the integration to production and then periodically after the integration is in production.

  1. In the navigation pane, click Observability, then Integrations.
  2. Hover over the row of an integration, and select View processing time statistics View processing time statistics.

    The Processing time panel opens. The panel shows the processing time for all integration instances for the integration.

  3. If needed, update the time filters at the top of the panel.
    The time filters appear at the top of the panel and to the right and include a drop-down filter and the ability to show data by the hour or by the day.

  4. Check whether the standard deviation varies from the average performance time.
    Ideally, the values for these lines should be similar. If they're not, a number of integration instances might take longer than expected to run.
    1. Above the first chart, select the following values to hide the lines for the values:
      • Minimum
      • Maximum
      Only the Average and Standard deviation values remain.
    2. Compare the Average and Standard deviation lines.

      Similar values indicate that the integration instances have consistent durations. When the lines aren't similar, consider investigating the outliers.

  5. Check for spikes in the the maximum and average performance times.

    Ideally, the values for the Maximum and Standard deviation lines should remain low over time. When these values spike, some integrations are taking longer than usual to run.

    1. Hide the Average line, and show the Maximum line.
      Only the Maximum and Standard deviation values remain.
    2. Review the Maximum and Standard deviation lines.

      Low values indicate that integration instances for this integration have similar processing times. On the other hand, spikes in the lines indicate that some instances have taken longer than usual to process. Consider investigating these spikes.

Find Integration Instances with Increased Processing Times

Charts and graphs help you quickly determine whether a slow-performing integration instance is an outlier or a trend.

  1. In the navigation pane, click Observability, then Integrations.
  2. Hover over the row of an integration, and select View processing time statistics View processing time statistics.
    The Processing time panel opens. The panel shows the processing time for all integration instances for the integration.
  3. Above the first chart, select the following values to hide the lines for the values:
    • Minimum
    • Maximum
    • Average
    • Standard deviation
  4. Above the first chart, select the following values to show the lines for the values:
    • 75
    • 95
    • 99
  5. Point to the spikes for the each of the percentile lines, and view their processing times in a popup.

    These numbers offer insights into the processing times of integrations. For example, consider the following chart.
    The time/date chart on the Processing time panel shows 3 lines. The 75th percentil line is low, indicating consistent processing speeds of around 3 seconds for 75 percent of integration instances. The 95 line has a small increase for a few days, indicating typical processing speeds of 3 seconds and higher processing speeds of around 15 seconds for a few days. The 99 line spikes above 2 minutes, 30 seconds for the 5th day, indicating a performance spike on that day.

    • The 75 line indicates that 75 percent of instances of this integration finished in about 3 seconds or less.

    • The 95 line indicates that 95 percent of instances of this integration finished in about 15 seconds or less.

    • The 99 line indicates that 99 percent of instances of this integration finished in about 2 minutes, 30 seconds or less.

  6. Determine whether the performance for the integration has remained consistent.

Measure the Running Times of an Integration's Actions

When an integration instance takes longer than expected to finish, determining the actions that took the most time can help you troubleshoot.

  1. In the navigation pane, click Observability, then Instances.
  2. Click Filter Filter icon, and update the filtering requirements for the page.
  3. Hover over the row of an integration instance, and select View details View icon.
    The activity stream opens.
  4. Review the duration for each part of the integration flow to determine which steps took the most time.

    The processing time appears after the timestamp, above the information about the activity.
    A gray box appears around the 97ms time, which appears on a milestone in the activity stream