The Service Level Policy defines conditions for measuring and reporting performance of a specific contract.

Note that this policy measures and reports on service level. It provides metrics, but it does not block any service calls.

In this policy, you can define:

Rule (one or more): Each rule represents a set of conditions you define to measure and report performance of a service contract. When a defined system condition matches a defined rule, an alert is raised. A Rule is composed of Rule Elements (Alert Code, Metric, Operator, Value, Units, and Interval) that you can configure to meet your service monitoring requirements.

Note: To see the alerts in Policy Manager: Policy Manager > Workbench. Alerts are not displayed in the Community Manager developer portal, but you can see the behavior with regard to whether messages are allowed or limited by attaching a Detailed Auditing policy to the API and reviewing the metrics and logs.

Access time: Defines the time periods during which the policy is enforced. You can define access time periods uniquely for each day of the week—all day, or a start/end range (HH:MM)—and the time zone.

Here's how it works:

The platform collects statistics for all API calls. The raw data is recorded in the database.

Platform processes run at intervals to analyze the data to see if any Service Level Policy limits have been reached.

The platform uses the time frame specified in the policy rule (see Interval below). For example:

A rule limit of 5 faults per hour is checked each hour, on the hour.

A limit of 100 transactions per 15 minutes is checked on the quarter-hour (for example, 1:00, 1:15, 1:30, 1:45).

Alert: If a rule has been violated for the specified interval, an alert is generated.

"Clearing" Alert: At the next scheduled interval, the rule is checked again. The platform applies the following logic:

If the message count is still exceeded, no action is taken; there is no additional alert, but the alert is not cleared.

If a rule is no longer exceeded, the Service Level Policy generates a "Clearing" alert to notify users that the violation is over.