[
https://issues.apache.org/jira/browse/ARIES-1394?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
John Ross updated ARIES-1394:
-----------------------------
Attachment:
UseExistingSubsystemResourceContentSetsRatherThanContentHeaderToIdentifyContent-after.cpu
UseCapabilitySetWithContentRepository-after.cpu
Screenshot-Java Monitor -
UseExistingSubsystemResourceContentSetsRatherThanContentHeaderToIdentifyContent.png
Screenshot-Java Monitor -
UseCapabilitySetWithContentRepository.png
Screenshot-Java Monitor - LocalRepositoryUsesCapabilitySet.png
Screenshot-Java Monitor - BasicSubsystemCacheRegion.png
LocalRepositoryUsesCapabilitySet-after.cpu
BasicSubsystemCacheRegion-before.cpu
BasicSubsystemCacheRegion-after.cpu
BigApplicationTest.java
AbstractPerformanceTest.java
> Provide more efficient implementations of the local and content repositories.
> -----------------------------------------------------------------------------
>
> Key: ARIES-1394
> URL: https://issues.apache.org/jira/browse/ARIES-1394
> Project: Aries
> Issue Type: Improvement
> Components: Subsystem
> Affects Versions: subsystem-2.0.2
> Reporter: John Ross
> Assignee: John Ross
> Labels: performance
> Attachments: AbstractPerformanceTest.java,
> BasicSubsystemCacheRegion-after.cpu, BasicSubsystemCacheRegion-before.cpu,
> BigApplicationTest.java, LocalRepositoryUsesCapabilitySet-after.cpu,
> Screenshot-Java Monitor - BasicSubsystemCacheRegion.png, Screenshot-Java
> Monitor - LocalRepositoryUsesCapabilitySet.png, Screenshot-Java Monitor -
> UseCapabilitySetWithContentRepository.png, Screenshot-Java Monitor -
> UseExistingSubsystemResourceContentSetsRatherThanContentHeaderToIdentifyContent.png,
> UseCapabilitySetWithContentRepository-after.cpu,
> UseExistingSubsystemResourceContentSetsRatherThanContentHeaderToIdentifyContent-after.cpu
>
>
> These measurements use a new test scenario to help flush out bottlenecks in
> the local and content repositories. The time it took to install 1 application
> was measured. The application consisted of 200 bundles. Half of the bundles
> each exported 50 packages. Half of the bundles each imported 50 packages.
> This gives a total of 5000 requirements and 5000 capabilities.
> Before
> Average time across 1 trials: 13915.0 ms
> Average time across 1 trials: 14628.0 ms
> Average time across 1 trials: 14196.0 ms
> Average time across 1 trials: 13905.0 ms
> Average time across 1 trials: 14021.0 ms
> 14133
> After
> Average time across 1 trials: 7781.0 ms
> Average time across 1 trials: 7782.0 ms
> Average time across 1 trials: 8085.0 ms
> Average time across 1 trials: 7850.0 ms
> Average time across 1 trials: 7971.0 ms
> 7893.8
> A 44.2% overall improvement.
> Before, obtaining the subsystem region took 2254 ms, or 14.4%. After caching
> the region context bundle and using it to perform the region lookup, the
> effort no longer registered on a profiler with a 50 ms sampling period.
> Before, finding providers from the local repository took 2260 ms, or 15.3%.
> After adding the capability set, it took 747 ms, or 5.9%.
> Before, asking whether or not a resource was content took 2152 ms, or 16.9%.
> After answering by using existing content resource sets rather than the
> content header, the effort no longer registered on a profiler with a 50 ms
> sampling period.
> Before, finding providers from the content repository took 4510 ms, or 28.8%.
> After adding the capability set, it took 49 ms, or 0.4%.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
