Introduction to the aggregation report of Jmeter test results analysis

aggregated reports (Aggregate report)

For each request, it counts the response information and provides the number of requests, average, maximum, minimum, error rate, approximate throughput (in Requests/sec), and throughput in kb/seconds.

The throughput is counted from the point of view of the sampled target (for example, the remote server set in the HTTP request sample). JMeter will take into account the total response time of the generated request, so if there are other samplers or timers in the same thread group, the total time will be increased, thereby reducing the throughput value. Therefore, two samplers with a different name only have a throughput of half the sum of the two sampler's throughput. Proper selection of the sampler name is important for obtaining the best results in the re-aggregation report.

Personal understanding: Regardless of whether there are other samplers or timers, here is mainly based on the calculation of time and number of requests, throughput = Number of requests/total time, take the timer, it has a thread delay function, without increasing the number of requests, increase the total time, natural throughput is reduced.

This author has done an experiment, each run, each time manually run, and do not empty the results before each run, run the test, view the aggregated report display. Because of the manual operation, every two runs, there are idle periods, in this period of idle time is sometimes longer, a few minutes to a few 10 minutes.

The results show that the aggregated report is cumulative, that is, the results of each run are counted based on the results of the previous run, including the number of requests that were initiated, such as the sum of the requested samples, such as I run 11:00, initiate 10 requests, run 11:20, launch 10 requests, At this point the aggregation report shows that the number of requests is 20, and the throughput at this time is far from the first run, and the individual guesses that it has counted the non-operational time from 11:00 to 11:20. Therefore, the total time is greatly increased.

Uncheck the label does not include the name (include group name in label) check box

Note: When using aggregated reports, do not use the same request sampler name in the test plan

Table Header Description:

Label-as pictured, uncheck "Include Group name in label?" check box is the name of the request sampler, otherwise "request Sampler Thread Group: Request Sampler Name"

# Samples-The number of requests sent with the same request sampler (note: This value is constantly accumulating). For example, if 10 threads are set to 10 and iterate 10 times, the value increases 10*10=100 every time you run a test.

Average-The average response time for a single request is by default, and when transaction Controller is used, the average response time can also be displayed in transaction units

Median-The median number. Represents 50% of the total number of sample requests that are less than the time value of the response time

90% line-represents 90% of the total number of sample requests with a response time not exceeding that time value

Min-Minimum response time to request samples for the same request sampler

Max-Request Sample maximum response time for the same request sampler

Error%-Percentage of requested samples with errors

Throughput-throughput is measured in terms of "requests/second, Requests/minute, Requests/hour". The time unit has been selected as second, so the display rate is at least 1.0, which is 1 requests per second. When the throughput is saved to the CVS file, the Requests/second is used, so 30.0 Requests/second is saved to 0.5 in CVS

Kb/sec-throughput as measured by kilobytes/seond

Attention:

1. No special instructions, the above time units are MS

2. Request response time refers to the entire time when a request is made from the client to be responded to, and then it's done, and then it's down. Jue born in the free life or not born