Investigate Poor Transaction Performance

Metric data, like the average response time of an important component, can inform you about the experience of customers using that component. However, this data does not help you understand cases where performance is unusually slow. When transactions are slow, a transaction trace can be something like an x-ray: showing details that are not visible from the surface, and letting you see where the transaction spent its time in a fine detail. Application Performance Management (originally Wily Introscope) pioneered the Transaction Trace approach to gaining deep insight into individual transactions. Transaction Trace monitors the activity of individual transactions as they flow through agent-monitored applications. While metric data tells you when there is a bottleneck in traffic, transaction traces can you tell you about the experience of a single car: where it was delayed, for how long, and even why. Transaction traces are stored, so you can view their details hours or days after the transactions first occurred.
apmdevops106
Metric data, like the average response time of an important component, can inform you about the experience of customers using that component. However, this data does not help you understand cases where performance is unusually slow. When transactions are slow, a transaction trace can be something like an x-ray: showing details that are not visible from the surface, and letting you see where the transaction spent its time in a fine detail. Application Performance Management (originally Wily Introscope) pioneered the Transaction Trace approach to gaining deep insight into individual transactions. Transaction Trace monitors the activity of individual transactions as they flow through agent-monitored applications. While metric data tells you when there is a bottleneck in traffic, transaction traces can you tell you about the experience of a single car: where it was delayed, for how long, and even why. Transaction traces are stored, so you can view their details hours or days after the transactions first occurred.
Find Traces of Slow or Failing Transactions
Experience View helps analysts find useful transaction traces to investigate. In the Business Transactions tab, analysts can then view transaction trace summaries and details. This information helps you understand transaction performance and solve poor performance by identifying when, where, and why performance degrades.
Follow these steps:
  1. In the left pane, click
    Experience View
    .
    The Experience View shows individual experience cards. Each card shows a summary. Red items indicate slow or failing transactions.
    Note:
    For more information about Experience View, see Monitor Performance Using Experience View.
    You can also start your investigation from the Map. In the left pane, click
    Map
    , and skip the next step. For more information about the Map, see View Component Relationships in the Map.
  2. Browse the cards and click the
    Notebook
    icon on a card of interest.
    The Analysis Notebook shows details about the experience. The Relationship Flow shows the transaction paths of the selected experiences. This map gives context for the event that occurred.
  3. Analyze the map and identify a node (component) that is flagged as a problem or anomaly.
    - A red circle indicates that the component is an actor in at least one problem or anomaly.- A concentric red circle indicates that the component is a culprit in at least one problem or anomaly.This component might be the source of performance degradation in your application environment.
  4. Click either individual nodes or groups of nodes for a maximum of 1,000 nodes. Select the
    Application Layer
    or the
    APM Infrastructure Layer
    in the map to see all the traces that are collected by host, agent, or application.
    A component chart appears and shows healthy and poor transactions.
    Click here for component chart help...
    The component chart shows a maximum of 20 nodes. Use component charts to compare live or historic metrics between arbitrary nodes. The component chart contains more metric information than is available for every component in the map. Use the switch options (dots) at the bottom of the chart to show the following information.
    • RESPONSE TIME histogram: The histogram shows the aggregated average response time per seconds for healthy and poor transactions. The histogram lets you easily compare data and identify trends, for example, when the number of traces increased or decreased.
    • AVERAGE RESPONSE TIME sparkline: The sparkline shows the general shape of the variation over time of the BlamePoint metrics for any component. The sparkline shows information for the active time period that is selected in the timeline. Mouse over any point on the sparkline to see a numeric value.
    • TRANSACTION VOLUME chart: The bar chart helps you quickly determine the level of transaction volume. Bars also provide easier identification of trends in volume.
    The component is highlighted in the
    AFFECTED APPLICATION COMPONENTS
    pane.
  5. Click the
    Business Transactions
    tab.
    A summary list shows traces that correspond to the component for the range that is selected in the timeline. Traces for inferred components, such as for backends, web services, or sockets are listed although these transactions are not monitored by agents. The traces show the duration times and are color-coded. Each color indicates a characteristic that is associated with a transaction, for example, red indicates an error. You can identify problematic methods by noticing the duration of the trace. Unexpectedly long traces are likely causes of slow transactions.
    Note:
    The list refreshes automatically when you use Live mode.
    The list shows the following trace information for a maximum of 2,000 traces.
    Url
    --The URL that was invoked to initiate this transaction, or the path to the component that initiated the transaction
    Name
    --The name of the high-level component, for example: Default
    Timestamp
    --The start time, in the system clock of agent host computer, of the invocation of the selected component
    Duration
    --The execution time in milliseconds of the selected component
    Trace Type
    –The type of trace: Error, Stall, or Other
    User Id
    --The ID of the logged-in user that is running the transaction
    This information helps you understand the sequence of calls within a time period and evaluate performance.
    Note:
    Not all the information is available about deep visibility components.
  6. Perform one or more actions:
    • Click the
      pop-out button
      pop-out button shows full view to open the viewer in a full-view dialog. Press the
      Esc
      key to close the dialog.
    • Select an option from the
      Trace Type
      drop-down list to categorize traces by characteristics. A trace can have more than one characteristic:
      Error
      : Returns transaction traces with an error characteristic. An error is an exception reported by JVM or HTTP error codes. For example, errors include an HTTP error status, an SQL exception, or a Java exception. If the Trace Type is Normal but an error triggered an automatic transaction trace, the component shows as an error and counts in the error total.  When a transaction trace includes an error snapshot, two transaction traces are listed:
      - One trace with duration
      - One trace with no duration (N/A), which shows the error snapshot
      Therefore, when a red transaction (error) includes an error snapshot, two transactions are listed whereas the Map shows a count of one.
      Stall
      : Returns transaction traces that have an error message of
      Stalled Transaction
      . A stall is a transaction, or component of a transaction, that has not completed within a specified time threshold. When a transaction trace includes a stall, two transaction traces are listed:
      - One trace with duration
      - One trace with no duration (N/A), which shows the stall.
      Therefore, two transactions are listed whereas the Map View shows a count of one.
      Other
      : Returns all transaction traces other than the traces with errors or stalls. Light orange differential analysis orange.png indicates that Differential Analysis triggered an alert--a transaction has uncontrolled variance.
    • Click a
      column header
      .
      The list sorts traces by the column type, for example, by
      User Id
      . Trace Type is not sortable by clicking the column header.
      Note:
      User ID is the identification of the logged-in user who is running the transaction (if it is configured and available).
  7. Examine individual components and trace data. You can look at call sequence and inside the code to determine the cause of a problem.
More Information: