Performance Monitoring

To monitor API performance and by inference, the performance of underlying systems, APs and TPPs should provide built-in performance metric gathering into their APIs. For APs, this information may be published in the Meta Directory under the Get Performance API call. If published, TPPs can query this information on a scheduled basis to determine which APs are operating optimally per time and may apply this intelligence to load scheduling and optimal path routing.

However, TPPs can make such decisions using metrics they maintain if not published by APs via APIs. Note that reporting requirements of this standard instruct that APs must provide monthly reports showcasing these performance metrics in daily summary and detailed formats. Alternatively, APs can provide performance monitoring dashboards which registered TPPs can access at any time to generate reports or locate patterns.

The following types of performance information are required from all APIs published by APs:

APs should record individual API performance metrics with at most 5 minute intervals between data points. Any interval missed is assumed to be downtime except within a period coinciding with a scheduled change or maintenance activity. For investigative purposes, these metrics can also be kept on a transactional level at the discretion of the AP. In addition, APs are required to provide API availability metrics. These metrics should be collected:

  1. From outside the APs' networks

  2. Over the internet

  3. Through at least two major Internet Service Provider routes.

  4. Per API endpoint e.g. one for branch related methods, another for ATM related methods etc.

The metrics give an external view of the organization's availability cutting across network, operating systems and application layers. External solutions or organizations can be engaged to provide and publish this data on behalf of APs.

Last updated