Performance metrics
Tingyun Server
Tingyun Server through deploying agent in application server, monitors application codes, relational database, non-relational database and external service, collects relevant performance data in real-time and does intelligent analysis, quickly identifies the bottlenecks of application performance, to provide performance diagnosis, tracing and optimization basis.
Application server response time
The time from the application server receiving the request to returning the response
Web application process
Web application process (Web Action) refers to the Web application to deal with a separate Web access request process. A complete Web application process is from the front-end browser initiating the request to the Web application outputting response.
Background task
It refers to a background process or task that runs and is detected in an application runtime environment where the application agent is deployed, but does not belong to any Web application process.
Apdex
Apdex defines three user satisfaction intervals based on the application response scores calculated for the applied ApdexT values. The three intervals of "satisfying", "tolerable”, "disappointed” are divided by the response time value “T”. T "to represent the response time limit or “ threshold "(Threshold) that the user satisfied with the application performance, that is, the first interval" satisfactory "bottom line, such as 0.5 seconds, the satisfying interval is 0 ~ 0.5 If the response time is longer than the T value, the user is somewhat dissatisfied. The next interval "tolerable” of the threshold is T and 4T, that is the tolerable interval is between 0.5 to 2 seconds; Users began to consider giving up the longer response time., The response time of the last interval "disappointed" is greater than 4T, that is, more than 2 seconds.
Apdex index
Apdex samples the tasks that occur in the application and divides the samples into the corresponding satisfaction interval, counts them, and calculates the Apdex index using a formula. Apdex index = [satisfaction quantity + (tolerable quantity / 2)] / total number of samples. In fact, the meaning of this formula is that a satisfactory sample score is 1, a tolerable sample score of 0.5, a disappointed sample score of 0, so the formula can also be written as:Apdex index = (1 x satisfied number + 0.5 x tolerable quantity) / total number of samples. In this way, the sampling result is quantized as a value between 0 and 1, that is, "Apdex index", 0 is not satisfied with any user, 1 on behalf of all users are satisfied.After the statistics, Apdex corresponding to the degree of user satisfaction, as shown in the figure below, the Apdex score for application performance is closely related to the user's experience, providing managers with a way to evaluate user satisfaction by application performance quantization.
Error rate
The number of errors or exceptions that occur per minute on the application server side as a percentage of the total number of requests
Throughput
The throughput of the application process, that is, the number of requests per minute
Response time and throughput
The average response time and the number of requests per minute for the current application
Block time
The time from the web server to the application server
Time offset
The call time difference of each layer of web application components
Thread profiling
Without affecting the user experience, collect thread state with a very low consumption. After the profiling is complete, it displays the code time consuming, to help customers real-time understand the web request, background tasks and other performance data
Wall clock time than ratio
Web application process time-consuming wall time ratio = The total time - consuming time of the Web application process at the time scale of the graph/Chart abscissa granularity time，the value reflects the busyness of the Web application process and the consumption of CPU resource.