Calculating an application performance score
The Application Performance Score object defines the application traffic that will be monitored and which application performance metrics to evaluate. It also provides application performance thresholds to be used in the evaluation.
For each metric, the observed traffic is compared against the threshold and is classified into one of three categories:
- Good — The baselineused by monitors and performance reports to establish a standard by which sunsequent performance can be measured for the application is good, which indicates that the application is performing within the expected levels (below the threshold). Users should be happy with application performance.
- Tolerated — The performance of the application is less than expected, but still performing within a range that users should be able to tolerate (between the threshold and four times the threshold).
- Frustrated — The application is performing poorly (more than four times the threshold). Users will be frustrated.
The number of good observations for all metrics with a threshold are totaled and given a full weighting; the number of tolerated observations for all metrics with a threshold are totaled and given a half weighting; and all frustrated observations are given a zero weighting. These weighted totals are summed and divided by the total observations.
aps = 10 * ((1 x number of satisfied samples) + (0.5 x number of tolerated samples) + (0 x number of frustrated samples)) / total samples
EXAMPLE
For HTTP, a threshold is configured for Network Delay as T = 300 msec and a threshold is configured for round-trip time (RTT) as T = 40 msec.
In one 10s period, 11 flows are sampled for HTTP with the following results:
- 2 flowthe network traffic between network objects samples have a network delay of > 1200 ms (frustrated samples)
- 3 flow samples have a network delay of > 300 ms but < 1200 ms (tolerated samples)
- 6 flow samples have a network delay of < 300 ms (satisfied samples)
- 1 flow sample has a RTT of > 40 ms but < 160 ms (tolerated samples)
- 10 flow samples have a RTT of < 40 ms (satisfied samples)
The APSApplication Perfromance Score score is calculated as follows:
aps = 10 * ( 1 * (6 + 10) + 0.5 * (3 + 1) + 0 * 2) / 22 = 8.1
Setting thresholds
The appropriate thresholds for an application is unique for each network environment. Thresholds can be set manually when configuring an APS object or the Exinda appliance can analyze the traffic for an application for a baseline period and create a recommended set of thresholds.
For more information refer to Configuring application performance score objects.