Automated Transaction Diagnostics

Splunk AppDynamics continuously collects Transaction Snapshots that capture all events in a transaction path for use in performance troubleshooting and root cause analysis. Although rules and limits exist to prevent excessive resource consumption, you may have hundreds of individual snapshots to sort through as you investigate a particular performance issue.

The Automated Transaction Diagnostics (ATD) capability allows you to be proactive rather than reactive for anomalous transaction behavior. ATD uses machine learning to capitalize on the collected snapshots by automatically recognizing anomalies not yet severe enough to trigger alerts but significant enough to cause future production issues. In seconds you have a prioritized list of causal factor transactions that you can use to manage problematic business transactions before they become visible production slowdowns or outages for your customers.

ATD's guided root-cause analysis clearly exposes the anomalies with the contributing tiers, exit calls, or inter-tier network issues. This allows a broader range of less specialized users to more accurately analyze the behavior and significantly reduce repair time.

Automated Transaction Diagnostics Workflow

Note: The Transaction Diagnostics panel becomes visible after you select a business transaction.

The diagnostic process detects performance issues, automatically identifies the contributing transactions, and provides a report that simplifies root cause analysis allowing you to fix problems before they become production issues.

  1. From the Home or Applications tab, choose the application you want to analyze.
  2. Select Business Transactions.
  3. Double-click a transaction to investigate.
  4. From the Dashboard tab, review Transaction Diagnostics in the right panel.
  5. Select a highlighted anomaly to open the Transaction Diagnostics details page.
The interactive graph consists of two charts. The upper graph displays transaction response times that correspond to the time range you choose in the lower navigation bar. The yellow plot bands highlight periods of abnormally slow response times.
  1. Use the slider in the lower navigation chart to select a time range from five minutes to eight days.
  2. Click any plot band in the navigation bar to view details for that time range. The chart automatically adapts to display the incidents for your time range choices and includes a baseline for the transaction.
  3. Review the Details of Performance Issue section to view the Suspected cause. To manually review snapshots from the anomalous period, review the lower Snapshots of Impacted Transactions list.
  4. Double-click a snapshot in the list to view the transaction details.