Aggregation Framework

Overview

As of SAP Focused Run 3.0 FP02 the new Aggregation Framework for system- and self-monitoring metrics of the Monitoring & Alerting Infrastructure (MAI) is available. It replaces the old framework which is available in System Analysis Aggregation.

The Aggregation Framework offers you the following benefits:

  • You can store hourly aggregated metric values of System and Self-Monitoring for a period that exceeds the default value of 28 days.
  • You can define which metrics on which objects shall be aggregated in a very fine granularity.
  • It improves the performance for dashboard and monitoring UIs displaying aggregated data, especially when displaying longer time frames (like daily or weekly aggregates over a longer period).
  • It extracts aggregated data using the OCC Dashboard interface in a performant way.

Within SAP Focused Run the new Aggregation Framework is used in the following apps:

Configuration

The prerequisite for configuring the Aggregation Framework is that the user has the role SAP_FRN_AAD_MOAL_REP_ALL (see New, Changed, and Obsolete Roles with SAP Focused Run 3.2).

Before starting the configuration, make sure that all MAI jobs are scheduled with default variant. To do so, execute transaction MAI_TOOLS, activate the expert mode, and in tab Administration, click Schedule all MAI jobs with default variant.

The configuration of the Aggregation Framework is done within app System Monitoring - Individual Maintenance and Assignment in the page Aggregation Framework (last page in the navigation bar on the left-hand side). 

 

It uses a variant concept. In each variant, you make the following settings:

  • Name and description of the variant.
  • How long the metric values shall be stored in hourly resolution (Retention Time). You can use the values Short, Medium, or Long. To set he length of these times in months, select Retention Time Configuration; the default values are as follows:

Retention TimeMonths
Short3
Medium12
Long24

  • The systems for which the metric values should be aggregated. You can specify the systems by name or via corresponding filter settings (e.g. all productive systems with lifecycle status Active).
  • The metrics that should be aggregated. As for systems you can specify the metrics by name or via corresponding filter settings (e.g. all availability metrics).
  • In the last step you can create a forecast showing which metrics on which objects are being collected via the variant.

 

The variant is immediately active after saving it. The aggregation is performed with a delay of 1-2 hours (e.g., when the aggregation job runs at 10.30 data from 8.00 to 9.00 of the same day will be aggregated).

For more information, see the in-app help of the Aggregation Framework.

Self-Monitoring and Self-Healing

In the Central Components of the Self-Monitoring Dashboard, two metrics are monitoring the status of the Aggregation Framework (the self-monitoring metrics are collected by job SAP_MAI_AGGR_SFM):

  • The Aggregation Status checks if the hourly aggregation is performed without any errors (job SAP_MAI_UDM_AGGR_HOURLY).
  • The Partitioning Status checks if the daily partitioning of table MAI_UDM_HOURLY is executed without any issue (job SAP_MAI_UDM_AGGR_PARTITIONING).


See the description of the metric if the rating is red. In case the aggregation fails, the next job run will perform the aggregation for missing time slots as well.

Migration from System Analysis – Configuration

The following information is only relevant for you if you have used the app System Analysis – Configuration for retention time of metrics used by OCC Dashboard and System Analysis.

Note that there is currently no migration of configuration or aggregated metric data from the System Analysis framework to the new Aggregation Framework. Therefore, configure the variants for the new framework so that no less data is aggregated than before. Use the forecast option for a review when creating the variants to find out which metrics on which objects are being collected via the variant.

System Analysis and OCC Dashboard have a logic implemented to load the data from the correct source: all data which is older than the activation of the Aggregation Framework is coming from the System Analysis framework, and all data which is newer is coming from the Aggregation Framework.

After activating the Aggregation Framework, deactivate the System Analysis framework manually. This is done by deactivating all variants in the app System Analysis – Configuration.

Known Limitations

The granularity for the aggregation is always on hourly level, there are no other options here. The maximum retention time in the Aggregation Framework is 24 months.

The aggregation is always done in UTC time. Since the granularity is always hourly, time zones with minute shifts compared to UTC are not supported. This is the case, for example, for the Indian time zone, which has a shift of 30 minutes. As a user in that time zone, you can use the aggregated data the applications, but single values are slightly incorrect. This is particularly relevant when displaying hourly aggregates. When displaying daily or weekly aggregates, the problem becomes significant less important. If data is required in such a time zone within a scope where raw data is available, this data will be used. That means correctness of the data is more important than performance. Therefore, the issue occurs only for data that is older than 28 days.