Creating a Federated Data Mart

A federated data mart enables you to use the data mart capabilities to perform analysis and monitoring for multiple deployments in your organization. You can consolidate ACM data from the data marts of multiple systems into one location, which enables you to compare system performance and availability across your organization. Each deployment still retains its own data mart, but the federated data mart enables you to analyze data across deployments. For example, you could set up a federated data mart to contain data from DEV/TEST/PROD deployments, or from multiple departmental deployments. The federated data mart is available following the third maintenance release of SAS 9.4.
Note: APM metric data is not included in a federated data mart.
To create a federated data mart, follow these steps:
  1. Verify that SAS Environment Manager is running on all of the deployments that you want to include in the federated data mart.
  2. Configure SAS Environment Management Service Architecture on all of the systems, and ensure that the ACM ETL is enabled. This step ensures that each deployment includes a data mart containing ACM data. Before attempting to implement the federated data mart, you should allow the data mart to operate on each deployment for several weeks in order to verify that the data marts are operating as expected and to understand the data that is being collected.
  3. Choose which deployment will host the federated data mart and which deployments will be contributors to the data mart. On the host deployment, run the fedInit script to create the directories and structure needed for the federated data mart. See Exporting Data Mart Contents By Using exportDataMart. The script also creates a server resource on the host deployment named Federated Data Mart. You can monitor this server resource as you would any other resource in SAS Environment Manager. See Monitoring Resources.
  4. To populate the federated data mart, perform a full export of all the data in the data marts of the contributor deployments. This ensures that you have a current view of all of the data in the data marts.
    To perform a full export, run the bin/exportDataMart script on each deployment. This script creates a .tar file in the Datamart/Exports directory.
  5. To enable the exported files to be imported by the host deployment, move all export files that you want to import to the emi-framework/LandingZone/FedDataMart directory on the host deployment.
  6. When the exported data has been copied to the directory on the host deployment, import the data into the federated data mart. To import the data, run either the bin/updateFedDataMart script or run the Load Data control action.
    The import process extracts data from the .tar files and transform the data so that the data from all of the contributor deployments can be merged. Although the data is merged, the federated data mart retains information that identifies the contributor deployments and adjust identifiers so that they remain unique across the deployments. After the data is imported, the federated data mart contains merged data for each contributor system.
  7. Whenever you want export additional data into the Federated Data Mart, use the Export Data Mart control action or run the bin/exportDataMart script on a contributor deployment.
  8. You can also set up an automatic export of data from a contributor deployment. To set up automatic export, modify the sasev.properties file (found in the Service Architecture root directory) and set the datamart.export.withETL property to true. When this property is set to true, the ACM ETL script automatically creates an export file with the latest data. This export file contains only the updates since the last time the ETL job was run. It is important to note the difference between this file and the export file that is generated by the manual export process. The manual export process creates an export file containing all of the data in the contributor deployment. The federated data mart's update processing automatically detects and handles either a full export or an incremental export.
    Regardless of whether you use a manual, full export or the automatic, incremental export, you must move the export files to the proper location on the federated data mart host machine. The automatic export process handles only the export process, and does not handle the movement of the export file from the individual contributor deployments. It also does not automate the loading of the new data into the federated data mart.
    Because the automatic export contains only incremental data, the export files that it creates are smaller than those generated by the manual export process. On the other hand, many more files are generated. In addition, the files are generated every time that the ACM ETL process runs (typically nightly), rather than only when you are ready to run the Import Data process on the federated data mart host machine.