[
https://issues.jboss.org/browse/JBTM-2180?page=com.atlassian.jira.plugin....
]
Michael Musgrove commented on JBTM-2180:
----------------------------------------
As a start on this task I have enabled the Oracle® JRockit Flight Recorder for a subset of
the performance tests in the perf CI job (
http://172.17.131.2/job/narayana-performance/).
Flight data is stored in the maven build directory (target/test-classes/*.jfr) and can be
read using the jmc tool available in the standard JDK (edition 7u40 and above).
What is the best way to evaluate other options given that there are many alternatives (and
trying out each one involves engineering work):
* JDK Collections (much of our code hasn't been updated to take advantage of newer
releases (Jesper has been looking into this);
* Goldman Sachs Collections;
* Google Collections (Guava);
* javolution;
* Highly Scalable Java;
* ...
Tests with flight data currently include:
Core:
com.hp.mwtests.ts.arjuna.performance.Performance1
com.hp.mwtests.ts.arjuna.performance.Performance2
com.hp.mwtests.ts.arjuna.performance.Performance3
com.hp.mwtests.ts.arjuna.performance.Performance4
JTA:
com.arjuna.ats.jta.xa.performance.OnePhase2PCPerformanceDefaultUnitTest
com.arjuna.ats.jta.xa.performance.OnePhase2PCPerformanceVolatileUnitTest
com.arjuna.ats.jta.xa.performance.OnePhasePerformanceDefaultUnitTest
com.arjuna.ats.jta.xa.performance.OnePhasePerformanceVolatileUnitTest
JTS
com.hp.mwtests.ts.jts.local.synchronizations.Performance
com.hp.mwtests.ts.jts.orbspecific.local.performance.Performance1
com.hp.mwtests.ts.jts.orbspecific.local.performance.Performance2
com.hp.mwtests.ts.jts.orbspecific.local.performance.Performance3
Determine whether or not we should use a specific collections
framework
-----------------------------------------------------------------------
Key: JBTM-2180
URL:
https://issues.jboss.org/browse/JBTM-2180
Project: JBoss Transaction Manager
Issue Type: Task
Security Level: Public(Everyone can see)
Components: JTA, JTS, Performance Testing, REST, Transaction Core
Affects Versions: 5.0.1
Reporter: Mark Little
Assignee: Michael Musgrove
We use many of the inbuilt collections implementations (such as Hashtable) which are well
known to be suboptimal in certain cases. Should we use a separate collections framework,
e.g.,
https://github.com/goldmansachs/gs-collections
There are pros and cons of doing this. But first we need to determine whether it really
makes sense from a performance perspective.
--
This message was sent by Atlassian JIRA
(v6.2.6#6264)