Prediction Analysis dashboard

As a an operator or site reliability engineer (SRE) user in BMC Helix AIOps, use this dashboard to view the prediction analysis for the last 90 days. This dashboard displays a pie chart showing the overall prediction severity count, a bar chart showing the occurrences of predictions events against each service, and  an overall prediction analysis summary for the selected number of days. By default, the dashboard displays information for all services for the last 15 days and it can be extended to a maximum number up to last 90 days period. 

Operators or SREs can use and analyze the past predictions to get better insights and do some of the following tasks:

  • Identify the predictions data of the top impacting issues or the top 5 impacted services in recent times
  • Proactively resolve such issues or services from future occurrences
  • See if there was a prediction event that they overlooked in the past
  • Set up an advance alert for specific issues or events.
  • Optimize the service to minimize service degradation

Related topics

Configuring-dashboards-with-additional-customizations

Creating-and-customizing-dashboards

Sharing-and-embedding-dashboards

Example: Analyze and get insights from past predictions

Jim, an operator in Apex Global wants to get insights from the past predictions analysis for the last 15 days to identify the most impacting issues or impacted services

Jim performs the following steps to get the required information:

  1. Jim logs in to BMC Helix Dashboards and opens the Prediction Analysis Dashboard.
  2. In the time range filter, he selects Last 15 days.
    The dashboard displays the relevant data. 
  3. (Optional) In the Service Name and Severity filters, he selects either one or all services and severity types.
    By default, All is selected in both Service Name and Severity filters.

Jim views the Prediction severity panel and the Prediction Occurrences by Services panel to see the number of critical and major events and the number of events against the top 5 impacted services. Jim views the Predictive Events table to check the additional details, such as event creation time, object class, metric name, and message content associated with the service event along with the severity and status details for each.

The following image shows the Prediction Analysis dashboard with sample data:

prediction_analysis_dashboard_252.png

To view the Prediction Analysis dashboard

View the Prediction Analysis dashboard in one of the following ways:

Viewing from BMC Helix Dashboards:

  1. Log in to BMC Helix Dashboards.
  2. From the navigation menu menu_icon.png, click Dashboards.
  3. In the Operations Management folder, click Prediction Analysis. 

Tip: Quick access from the home page

To quickly open the dashboard from the home page, mark it as a favorite by using the star icon. Additionally, after you open a dashboard, it is available under Recently viewed dashboards on the home page.

Panels in the Prediction Analysis dashboard

The following table describes the panels in the Prediction Analysis dashboard:

Panel

Description

Example

Dashboard filter

By default, the dashboard displays the data for the last 15 days. You can filter the data by using the time range global filter for up to last 90 days.

global_time_filter.png

Service filters

Service Name: Lists up to 1000 service names.

Note: Only 1000 service names are listed. 

service_name_filter.png

Severity: Lists all the severity types.

service_severity_filter.png

Prediction severity

Displays the number of critical, major, and minor predictive events.

Prediction_Severity_panel.png

Prediction Occurrences by Services

Displays the number of predictive events for the top 5 impacted services.

Prediction_occurrences_panel.png

Predictive Events

Lists the individual predictive events from each service with the following details:

Creation time: The date and time stamp of the event creation.

Service name: Name associated with the service created in BMC Helix AIOps or discovered in BMC Helix Discovery.

Host name: The CI with which the metric is associated. 

Object class: Identifies the class of an object. An object can be a sub-component of the host to which the event is related. For example, it could be the name of the disk on which the predictive event is reporting the problem.

Metric name: Name of the metric that triggered the predictive event.

Metric value: The data point from which the forecasted values violate the threshold value before an event is generated.

Message: A brief predictive event message text.

Threshold value: The value that is considered for generating the prediction event. The forecasted data must exceed or fall below this value to trigger the prediction.

Severity: Indicates the severity type.

Status: Indicates whether the event was open or closed.

pr_service_key: Service key for which the prediction event is generated. 

predictive_events_252.png

 

Tip: For faster searching, add an asterisk to the end of your partial query. Example: cert*