commit 1b9fbb02ac2d2eb402c18877079d1bdbae03a620 parent 743cf9ed4559f27a7fd21c314a69cb1d64b7d6af Author: MS <ms@taler.net> Date: Sun, 9 Apr 2023 18:25:02 +0200 comment Diffstat:
| M | nexus/src/test/kotlin/SchedulingTest.kt | | | 11 | +++++++++++ |
1 file changed, 11 insertions(+), 0 deletions(-)
diff --git a/nexus/src/test/kotlin/SchedulingTest.kt b/nexus/src/test/kotlin/SchedulingTest.kt @@ -21,6 +21,17 @@ import kotlin.text.get * give the chance to monitor the CPU usage with TOP(1) */ +/** + * It emerged that whether asking transactions via EBICS or x-libeufin-bank + * is NOT performance relevant! For example, asking for a bank account + * balance - via the plain Access API - brings the CPU usage to > 10%. Asking + * for /config - via Integration API - used to oscillate the CPU usage + * between 3 and 10%. + * + * The scheduler's loop style is not relevant either: a while-true & delay(1000) + * or a Java Timer did NOT change the perf. + */ + // This class focuses on the perf. of Nexus scheduling. class SchedulingTest { // Launching the scheduler to measure its perf with TOP(1)