Contact Us
How you can contact us:

System Anomaly Prediction

System Anomaly Prediction in SAP Focused Run helps system administrators to predict anomalous situations within a system in the system landscape with the help of prediction models. The System Administrator can then use the tools available in SAP Focused Run to do a root cause analysis and avoid the situation from happening in the managed system. The data collected from the managed systems is used by the prediction models to predict anomalous situations.

If you have SAP Focused Run 4.0 SP00 or lower: Please refer to the System Anomaly Prediction - until SAP Focused Run 4.0 SP00 page.


  • Prediction
    Shows risk score for the systems configured for prediction. Provides a detailed view of the models and metrics used for calculating the risk score.
  • Settings
    Simple to use Settings page to activate / de-activate systems and model versions
  • Alerting
    Provides alerts for anomalous situations once the risk score exceeds the risk score threshold set for the system and model combination in the settings.
  • Custom Model
    Custom built models can be integrated with System Anomaly Prediction through the approach of BAdI

Target Group

The target group are service providers and customers who wants to avoid unavailability and bad performance situations from happening in the systems managed by SAP Focused Run.

Models Available

The below notes contain the information about the models and provides the model files that needs to be used for different support packages of SAP Focused Run.

ReleaseRelevant Model Information
SAP Focused Run 4.0 SP01SAP Note 3338259 - System Anomaly Prediction in SAP Focused Run 4.0 FP01 - Model Definitions and Model Files for Predictive applications

Configure Anomaly Engine and Model Deployment

The first step towards infrastructure preparation for System Anomaly Prediction is to implement the SAP Focused Run Master Guide completely.

The Machine Learning Platform ‘ML Service from SAP Cloud ALM' requires you to have a Cloud ALM tenant. Thus, before proceeding with the configurations in the SAP Focused Run – Cloud Service Management, kindly get request a SAP Cloud ALM tenant if you don't have one.


Configure Anomaly Engine and Model Deployment

  1. Log in to your SAP Focused Run ABAP system on the production client.
  2. Start transaction SM30.
  3. Enter as table name: PAS_SM_GEN_CONFI.
  4. Choose Maintain.
  5. Choose New Entries and provide the following values:
  • Param Count: 1
  • Param Value: 30 (Number of days, for which the prediction data is retained)

6  Choose Save.

7  Choose New Entries and provide the following values:

  • Param Name: ML_PLATFORM
  • Param Count: 1
  • Param Value: CALMENDPOINT

8  Choose Save.

9  Download the latest model definition from SAP Note 2706779.

10  Upload to SAP Focused Run the latest model definition (.zip file), by running the report PAS_SA_IMPORT_MODEL using transaction SA38.

Roles and Authorizations

The configuration can be done in the System Monitoring application only.

1  System Anomaly Prediction Configuration

SAP_FRN_AAD_MOAL_ALL - All authorizations for System Monitoring & Alert Management Administration/Configuration

2  System Anomaly Prediction Display

SAP_FRN_APP_MOAL_DISP - Display authorizations for System Monitoring & Alert Management

SAP_FRN_APP_MOAL_ALL - All authorizations for System Monitoring & Alert Management

3  System Analysis:

System Anomaly Prediction Display

SAP_FRN_APP_SYA_ALL - All authorizations for the System Analysis application (end user)

Schedule Jobs

Schedule the following background jobs, using transaction SM36, and select as Target, the previously created Job Server Group FRN_JOB_PUBLIC. These jobs cannot be scheduled via task list and must be scheduled manually.


Job NameDefine Step UserABAP Program NameStart TimePeriod Value




Configure the SAP Cloud ALM based ML Platform

The ML Platform for System Anomaly Prediction can be changed very easily from R Server to SAP Cloud ALM by following the given steps.

Configuration in Cloud Service Management

Open the Cloud Service Management in SAP Focused Run.

2  Click the ‘Add' button, and import ‘SAP Cloud ALM' in the cloud service type field. Specify all the details and click on Save button.

NOTE: Currently, only OAUTH based authentication is supported by SAP Cloud ALM service type.

The connection status of the endpoint can be seen in the Cloud Service as shown below.

Configuration in System Anomaly Prediction – Expert View

1  Go inside the System Monitoring application and open the System Anomaly Prediction application.

2  Click on the settings button and drill down the System Anomaly Prediction setting. Click on Expert View.

3  Click on the ‘General Settings' tab and choose ‘ML Service from SAP Cloud ALM' as the Machine Learning platform. Choose the External Service ID for each customer network. Click on Save button.

The connection status of the HTTP Endpoint for each customer network configured is available as well.

NOTE: All the Cloud services that are created in Cloud Service Management are auto populated in the SAP Cloud ALM Account Mapping.

Completion of these steps marks the finished migration of the ML platform from R-server to SAP Cloud ALM based ML Service.

Personal Data

Personal Data Identification and Deletion in Scenario of System Anomaly Prediction

Predictive applications store the user ID in the following tables;


If you want to check whether personal data is stored in the application, you can execute the report PAS_PERS_DATA_USAGE.

Personal data that is stored in the application can be deleted by running the report PAS_PERS_DATA_DELETE.

The execution of the above-mentioned reports is logged in SLG1 using object “PAS”.


Support Components

You can raise your incidents in the support component SV-FRN-APP-SYM.


System Monitoring

Navigation to the page is possible from “Status Overview” view available in the Overview page or by directly selecting the “System Anomaly Prediction” page. The relevant display authorizations as defined in Roles & Authorization needs to be assigned to the user.

System Anomaly Prediction - Overview & Details

The overview section shows the different system bucketed into the categories of;

  • Predicted Critical,
  • Predicted Okay,
  • Not Configured and
  • Insufficient Data.

The details section shows;

  • Extended System ID along with System Type,
  • Customer Network,
  • Current Risk Score (hover over to see the actual value and the threshold),
  • Risk History (worst) across model versions of the system in mini chart (last six hours) and
  • Risk History per model version and of longer duration in the pop-view.

System Anomaly Prediction – Models View

The models view shows the;

  • Extended System ID along with system type,
  • Model Name along with Model Version,
  • Risk History of the specific model version in mini chart (last six hours) and
  • Risk History of the model version and of longer duration in the pop-view.
  • Current Risk Score (hover over to see the actual value and the threshold),

The metrics view shows the;

  • Extended System ID along with the System Type
  • Model Name along with Model Version
  • Metric Name along with Context Name
  • Current Metric Value
  • Metric History in mini chart (last six hours)
  • Link to Metric Monitor

System Anomaly Prediction – Metric Monitoring View

The metric history is shown along with the forecast value.

System Analysis

The Predicted Critical Systems are shown in System Analysis Application.

Selecting the icon, launches “System Anomaly Prediction - Details” with the predicted critical systems.


1.        What is risk score?

Risk score depicts the possibility of the occurrence of an anomalous situation. Higher the risk scores higher the chances of the happening of the anomalous situation. Risk score is calculated in real time when relevant metric data (specific to each model) from System Monitoring is run against the trained model.

2.        Is there an additional requirement for using ‘ML Service from Cloud ALM' for System Anomaly Prediction?

Yes. Kindly refer the section 4 “Setup” for more information about this. If the need is only for forecasting, then R server is not required.

3.        Is the data collected separately for Metric Forecasting and System Anomaly Prediction?

The data pushed for monitoring is used for forecasting and anomaly prediction. There is no separate data collection for these scenarios currently.