Managing Performance Thresholds › How Performance Thresholds Work › How Application Performance is Rated

How Application Performance is Rated

The management console rates the performance of an application by observing its TCP transactions and calculating:

• Network metrics for each client network that communicates with the application. If the 5-minute average for a Network metric exceeds the threshold, and the management console observed the metric the minimum number of times, the management console rates the corresponding 5-minute interval for the client network as Minor (yellow) or Major (orange) and creates a network incident.
• Server metrics for each server that hosts the application. If the 5-minute average for a Server metric exceeds the threshold, and the management console observed the metric the minimum number of times, the management console rates the corresponding 5-minute interval for the server as Minor (yellow) or Major (orange) and creates a server incident.
• Combined metrics for the application itself, which include both network and server metrics. If the 5-minute average for a Combined metric exceeds the threshold, and the management console observed the metric the minimum number of times, the management console rates the corresponding 5-minute interval for the application as Minor (yellow) or Major (orange).

  Note that the management console does not create application incidents. However, because Combined metrics include both Network and Server metrics, the management console can rate a server or network as Minor (yellow) or Major (orange), and rate the corresponding performance impact on the application itself. For example, if a Server metric degrades, the management console can also rate a Combined metric for the application as Minor.

To rate performance data as Normal, Minor (yellow), or Major (orange), the management console must collect 2 full business days of data, counting a business day from GMT midnight to midnight. For example, if the management console begins collecting data for a TCP session between a server port and client network on Monday at 3:30 p.m. EST, the management console cannot rate the performance of the application on that network until 7:00 p.m. EST on Wednesday. If the management console has not collected 2 full business days of data, the management console rates the TCP sessions between the server port and client network as Unrated.

By comparison:

• Baselines, which are available from the Explore button on the Operations page, report the historical norm for all TCP sessions between an application port on a server, and a client network. The management console calculates hourly baselines between all application ports, servers, and client networks, and tracks them for day of week, day of month, and activity of the past week. The management console uses the baselines to indicate when a performance condition is normal for that hour of the day. The management console requires 2 full business days of data to calculate baselines.
• Operational Level Agreement (OLA) reports, which are available from the Management page, quantify current performance and performance trends. Performance OLAs show how well an application is performing by counting, on an hourly basis, the percentage of transactions that are faster than a particular threshold.

More information:

Managing Application Performance OLAs

Combined Metrics

Network Metrics

Server Metrics