[
https://issues.apache.org/jira/browse/GEODE-3148?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
nabarun updated GEODE-3148:
---------------------------
Description:
After profiling the performance of queries such as SELECT DISTINCT e.id,
e.value FROM /region e, we could see that the bottleneck was caused by rehash /
resizing operation of the ObjectOpenCustomHashSet while adding new entries to
the data structure.
After working with struct set instead , resulted in performance improvements,
but there were few issues we need to fix with intermediate field names.
was:
After profiling the performance of queries such as SELECT DISTINCT e.id,
e.value FROM /region e, we could see that the bottleneck was caused by rehash /
resizing operation of the ObjectOpenCustomHashSet while adding new entries to
the data structure.
We experimented with ConcurrentSkipListSet and saw that performance was
improved. ConcurrentSkipListSet should be incorporated into Apache Geode if it
passes all the testing.
> Replace ObjectOpenCustomHashSet with alternate data structure
> -------------------------------------------------------------
>
> Key: GEODE-3148
> URL: https://issues.apache.org/jira/browse/GEODE-3148
> Project: Geode
> Issue Type: Bug
> Components: querying
> Reporter: nabarun
>
> After profiling the performance of queries such as SELECT DISTINCT e.id,
> e.value FROM /region e, we could see that the bottleneck was caused by rehash
> / resizing operation of the ObjectOpenCustomHashSet while adding new entries
> to the data structure.
> After working with struct set instead , resulted in performance improvements,
> but there were few issues we need to fix with intermediate field names.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
