Administration Guide › Aggregation and Correlation Engines (ACE1 and ACE2) › Calculation Engine Overview

Calculation Engine Overview

The following list demonstrates the Engine Workflow.

• Customer events enter the system through adapters.
• Adapter listeners collect events. The data is normalized and is easy to use. Adapter listeners enter this normalized data into the T_RAW_DATA table.
• The process of calculating the data is based on an event-driven system that runs on business logic created by the client in VBS script. Calculation values are kept in global variables in the business logic, which are updates for each event that occurs in the system. Examples for such events are the raw data entries, period start and end, resource structure changes, and timeslot start and end. The result for each calculated period is then defined in the specialized ‘Result’ function.
• (Optional) The system calculates raw data, converting it to intermediate data, which is placed into the T_INTERMEDIATE_DATE table. Intermediate data can be used for calculating other metrics. In this way, the calculations are only performed. All other metrics that require the same data utilize the results instead of performing redundant calculations.

Continuous Calculation between contract versions means that data used by business logic to calculate service level is automatically accessible to subsequent contract versions. This feature enables the SLA-manager to create new contract versions with new effective dates. This feature enables the retention of the contents of global variables. This feature allows the calculation to continue and enables previous values of global variables to be accessible to new contract versions.

The engine handles events in the order in which they arrive. The engine cannot undo the handling of events to go back in time to a previous calculation. The engine only takes the situation as it is now and handles new events as they arrive. In cases where there is a change at a timestamp that was already calculated, the engine recalculates. The engine keeps snapshots of the calculation states periodically as defined in the system configuration. These snapshots prevent the need to recalculate from the beginning of the contract version. The calculation starts from a known calculation state, and the simplest starting point is where it began. These snapshots are referred to as states. When the engine recalculates, it starts from the time of a state nearest to the time of the change.

When a metric requires recalculation, the engine searches for the closest state before the moment when the change occurred, and calculates from that state onward. For example, there is a scenario in which a change took place two weeks ago, and the closest state before it is three weeks ago. In this case, the engine calculates from the state three weeks ago. If no such state is found, the engine recalculates from the beginning of the contract version.