-
Bug
-
Resolution: Won't Fix
-
P4
-
6
-
generic
-
generic
In SQE performance test suite,
a big difference appears on the connector client methods
between rmi (jrmp or iiop) and jmxmp.
The test performs the following actions:
- create 500 connectors client
- connect the 500 connectors client to 1 server
- close the 500 connectors client
The creation phase is quite similar for rmi/jrmp, rmi/iiop and jmxmp:
~900 millis on ultra 60
The connect phase takes 2 more times using jmxmp:
4 seconds for rmi/jrmp, 2 seconds for rmi/iiop and 10 seconds for jmxmp
But the worst result concerns the close method where jmxmp is ~ 6000 to 7000
times slower than the 2 other protocols:
~800 millis for rmi/jrmp and rmi/iiop and 54 seconds for jmxmp !!
###@###.### 2005-1-13 16:09:29 GMT
a big difference appears on the connector client methods
between rmi (jrmp or iiop) and jmxmp.
The test performs the following actions:
- create 500 connectors client
- connect the 500 connectors client to 1 server
- close the 500 connectors client
The creation phase is quite similar for rmi/jrmp, rmi/iiop and jmxmp:
~900 millis on ultra 60
The connect phase takes 2 more times using jmxmp:
4 seconds for rmi/jrmp, 2 seconds for rmi/iiop and 10 seconds for jmxmp
But the worst result concerns the close method where jmxmp is ~ 6000 to 7000
times slower than the 2 other protocols:
~800 millis for rmi/jrmp and rmi/iiop and 54 seconds for jmxmp !!
###@###.### 2005-1-13 16:09:29 GMT