One of the secrets to anticipating any possible operational issues or the growth of a transactional system is to have a strong collection of statistical data. There is no doubt about this. The real question is, how can we get useful statistics KPIs for operational platforms?

Statistics are mostly based on analysis and trending KPIs. These performance indicators depend on and must be identified according to the platform that will be measured and monitored. In transactional systems, these KPIs can be grouped into two important groups.

Types of Statistics KPI for Operational Platforms

Statistics Health KPIs

Industry standards that represent physical or virtual resource states within the context where the applications are executed. Examples of these are:

• The number of cores used
• RAM or Swap memory consumption
• Access to storage
• Evolution of available disk storage
• Others

Statistics Business KPIs

In contrast, these KPIs are not always standard. They have a strong relationship with the product or platform to be measured. For example, for provisioning systems with several interfaces to external network elements, there will be presented:

• The number of incoming transactions differentiated by type: if they showed an error and it’s a completion status.
• Internal and external processing times: the overhead or internal processing time can be calculated.

Graphical Dashboard

Beyond the importance of their historical data storage and that they establish trends, these KPIs can be presented in dashboards. This structure help monitoring teams have an “at a glance”diagnosis of each task status. Also, the platforms’ state with which the monitored system interacts can be seen.

This type of monitoring platforms is useful in operational departments such as the NOC (Network Operation Center) type departments.

Alarms Based on Comparisons

As we presented before, one of the keys to obtaining trends is KPI samples storage. This stored information can be analyzed to obtain trends based on the generation of averages.

Those averages compared with real-time data may trigger alarms when a defined threshold is crossed.

Customer’s NMS Integration

To close the circuit of collection, analysis, reports generation and alarms, the monitoring system can count on the particularity of being able to be integrated or send notifications to NMS systems.

This integration is usually solved through SNMP and will allow this solution to send Health, Business or alarms information to the different corporate monitoring systems.

By using this guideline, a useful KPI definition can be made to monitor any transactional platform.