The performance test for sending 1000 AlarmMessages through the default message bus in the test environment configuration. It has been run several times in a row, and the following table shows the results (time measure between sending and receiving a message in milliseconds).
Tests where the same destination is used:
Test number |
Minimum |
Maximum |
Average |
Median |
Standard deviation |
---|---|---|---|---|---|
1 |
355 |
1156 |
385 |
372 |
1.5299588229753112 |
2 |
529 |
1326 |
564 |
550 |
1.4363526725703546 |
3 |
702 |
1136 |
743 |
727 |
1.3247962107433733 |
4 * |
868 |
73664 |
1020 |
912 |
72.75057295994307 |
5 |
1054 |
1653 |
1136 |
1104 |
2.268992287338148 |
6 |
1222 |
1957 |
1323 |
1286 |
2.5522053208940694 |
7 |
1396 |
3209 |
1553 |
1520 |
5.63892613890269 |
8 |
1564 |
3300 |
1709 |
1642 |
5.027059279539082 |
9 |
1747 |
6450 |
1981 |
2127 |
17.550838270578417 |
10 |
1936 |
7629 |
2683 |
2316 |
30.599870473582072 |
* Had to reconnect to messagebus during test, due to inactivity.
Test where different destinations are used for each test:
Test number |
Minimum |
Maximum |
Average |
Median |
Standard deviation |
---|---|---|---|---|---|
1 |
353 |
1100 |
378 |
369 |
1.2103449921406706 |
2 |
350 |
769 |
376 |
367 |
0.949808928153447 |
3 |
348 |
620 |
373 |
365 |
0.8967786794967865 |
4 |
345 |
650 |
373 |
364 |
0.9101142785386899 |
5* |
349 |
139739911 |
140114 |
364 |
139669.64911575476 |
6 |
341 |
718 |
384 |
368 |
1.2978066111713256 |
7 |
338 |
663 |
376 |
366 |
1.1685272782438585 |
8 |
338 |
706 |
375 |
360 |
1.2905522073903093 |
9 |
335 |
496 |
363 |
352 |
1.087482873428359 |
10 |
333 |
501 |
363 |
351 |
1.1523649595505758 |
* Had to reconnect to messagebus during test, due to inactivity.
3 Comments
Mikis Seth Sørensen
4 *: Seems like a error, why is the client disconnected from the message bus because of inactivity during a stress test??
Mikis Seth Sørensen
Please specify which version of the reference code the tests where performed on. This can be f.ex. done by tagging the git version.
Mikis Seth Sørensen
The delay statistics appear to increase between the runs. This is a serious problem, eg. the message bus performance degrades over time??