-
Bug
-
Resolution: Done
-
Blocker
-
8.0.0.GA-CR1, 8.0.0.GA
-
None
We observed a degradation in performance in the scenarios listed underneath;
We observed that the degradation got worse after JGroups was upgraded from 4.2 to 5.2;
The limits are the thresholds that were used in EAP7 tests: exceeding those limits means we have a degradation with respect to EAP7;
The tests are run on bare-metal machines to have comparable results (same machines used in EAP7 tests)
ejb + servlet scenario
- eap-8.x-clustering-stress-ejbservlet-repl/229 JGroups 4.2: RESPONSE TIME 39.4 msec EXCEEDED 3.3 LIMIT
- eap-8.x-clustering-stress-ejbservlet-repl/228 JGroups 5.2: RESPONSE TIME 81.1 msec EXCEEDED 3.3 LIMIT
distributed granular cache scenario
- eap-8.x-clustering-stress-granular-dist/190 JGroups 4.2: OK
- eap-8.x-clustering-stress-granular-dist/188 JGroups 5.2: RESPONSE TIME 7.9 msec EXCEEDED 4.4 LIMIT
replicated granular cache scenario
- eap-8.x-clustering-stress-granular-repl/86 JGroups 4.2: OK
- eap-8.x-clustering-stress-granular-repl/85 JGroups 5.2: RESPONSE TIME 11.0 msec EXCEEDED 5.5 LIMIT
heavyread scenario
- eap-8.x-clustering-stress-heavyread-session-dist/158 JGroups 4.2: RESPONSE TIME 96.7 msec EXCEEDED 24.2 LIMIT
- eap-8.x-clustering-stress-heavyread-session-dist/157 JGroups 5.2: RESPONSE TIME 55.8 msec EXCEEDED 24.2 LIMIT
distributed cache scenario
- eap-8.x-clustering-stress-session-dist/191 JGroups 4.2: OK
- eap-8.x-clustering-stress-session-dist/190 JGroups 5.2: RESPONSE TIME 6.7 msec EXCEEDED 4.4 LIMIT
- is caused by
-
WFLY-18066 ByteBufferMarshalledValue generates duplicate buffers during a single marshalling operation
- Closed
-
WFLY-18069 Eliminate unnecessary buffer copy when writing an object with known size via ProtoStream
- Closed