What is App Score?
App Score is a holistic health score for an application represented as a color coded letter grade. App Score is calculated for each application and environment combination. This allows users to compare scores for the same application across all environments where it is running. App Score represents an objective measurement and provides at-a-glance feedback of application quality and health.
How does it work?
Every hour Retrace will compute a score for all application/environments that are setup for Application Performance Monitoring (APM). The score takes into account satisfaction, performance, error and server metrics for the previous 7 days (168 hours) to generate the rating. Much more than Apdex (Satisfaction) alone, Retrace takes advantage of additional data provided by its agent to produce a holistic rating of application quality. The formula for the score will remain proprietary but the categories of data have been enumerated in this help document. Namely Apdex (Satisfaction), Error Metrics, and Server Metrics. At this time, the algorithm cannot be altered. The category weightings is the same for all applications and customers.
What do the grades mean?
Your application has scored very well!
This is the highest rating an application can achieve. In order to be ranked this highly an app must have score well in all categories.
Your application has scored well.
As the second highest rating your application is better than average but there are improvements that can be made.
Your application has scored as average.
An acceptable rating but improvements can be made, likely in multiple categories.
Your application has scored poorly.
This indicates that your application is experiencing poor overall health and is likely disappointing consumers.
Your application has scored very poorly.
This indicates that your application is experiencing very poor performance and is probably disappointing consumers. Corrective action should be taken to determine root cause(s) and address the issue(s).
App Score cannot be calculated for this application.
Since APM data is required in the scoring algorithm, this rating applies to applications that are not sending APM data.
App Score will be calculated when enough time has passed.
The algorithm requires at least 24 hours of data before a score can be accurately computed. Newly discovered applications will not have enough historical data to immediately compute the score and will therefore display this icon.
How Can I Improve My App Score?
Below are the suggested ways to improve your App Score for applications:
Correct slow running requests
Slow requests negatively affect Apdex (Satisfaction) score and are taken into account in the App Score algorithm. When deciding which request(s) to target for improvement, it is wise to consider the volume of calls a request receives as well as the average time it takes to process. A small improvement in a call with very high volume, will move the needle much more than a large improvement on a request that happens once an hour.
Correct slow running queries
Slow running queries are often the cause for the slow requests mentioned above. The same rule applies to queries as with requests, tuning those with a high call volume provides the most impact.
Provide Retrace with error logs
Since error counts and volume affects the App Score, failure to provide this data will be negatively reflected in your score. App Score is only one advantage to providing errors and logs to retrace. For more information about errors and logs click here.
Reduce the overall count of errors being thrown
A high error count will negatively affect your App Score for obvious reasons. An unhandled error could represent a failed user operation, display an error page to your consumer or even crash the entire application. Investigating and reducing the error volume will improve all aspects of your applications quality.
Reduce the count of new and regressed errors being created
Of particular concern are new and regressed errors. New errors represent something that hasn’t occurred until now and can be an indicator of sloppy coding, poor testing or data anomalies. Regressed errors are especially bothersome since they have recurred after being marked as fixed. Both of these types of errors can be reduced by writing unit tests for corrected defects at the time the resolution is coded.
Tune your application’s CPU usage
Inefficient applications have a tendency to consume large amounts of CPU while producing nominal results. If your application is routinely taxing the CPU, it is more likely than not causing other side affects that reduce application satisfaction.
Tune your Satisfaction Goal
If there are applications or individual requests in your application that are known and expected to be slower than others, the Satisfaction Goal can be modified to reflect the more realistic target response time. Making this goal more forgiving will increase Satisfaction and can affect App Score.