Java Magazine, Sept/Oct 2018
NUMBER OF REQUESTS 500 500 500 PRICE SERVICE MEAN PROCESSING TIME MS 100 1000 1000 PRICE SERVICE OUT OF SERVICE RATE 0 0 15 PRICE SERVICE STANDARD DEVIATION PROCESSING TIME MS 20 200 200 TIME TAKEN FOR TESTS SECONDS 531 50666 51052 NUMBER OF FAILED REQUESTS 0 0 85 REQUESTS PER SECOND 9400 987 980 HOLDINGS SERVICE RESPONSE TIME 50TH PERCENTILE MS 105 1011 1010 HOLDINGS SERVICE RESPONSE TIME 95TH PERCENTILE MS 138 1321 1334 ORACLE COM JAVAMAGAZINE SEPTEMBER OCTOBER 2018 42 the leading edge Lets test the application by simulating a total of 500 requests by 10 clients first when the price service takes 100 milliseconds ms to respond test 1 then when the price service takes 1000 ms to respond test 2 and finally when the price service takes 1000 ms to retrieve the request with a 15 out of service rate test 3 Here is an example of how you would run the test ab n 500 c 10 http localhost 8088 getHoldings Jake Table 1 shows the results from these tests As expected the time for the holdings service to process requests depends directly on the amount of time it takes the underlying price microservice to retrieve results plus some small added overhead Once I set the out of service rate some requests failed just slightly over the expected failure rate of 15 Because the holdings service processing time and error rate depends directly on the price microservice this implementation is not resilient to the behavior of the underlying subsystem test 1 test 2 test 3 Table 1 Results of testing the app with three runs of ab
You must have JavaScript enabled to view digital editions.