[jira] Updated: (UIMA-1200) Optimize UIMA AS XMI Cas serialization by reusing an instance of a XML Parser

["Jerry Cwiklik (JIRA)" <ui...@incubator.apache.org>]

     [ https://issues.apache.org/jira/browse/UIMA-1200?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jerry Cwiklik updated UIMA-1200:
--------------------------------

    Attachment: uimaj-as-jms-UIMA-1200-patch.txt
                uimaj-as-core-UIMA-1200-patch.txt
                uimaj-as-activemq-UIMA-1200-patch.txt

Optimized UIMASerializer to reuse XMIReader object when deserializing CAS from XMI. Each processing thread owns its own XMLReader that is pooled. Each instance is lazily instantiated on the first CAS deserialization from XMI.  The pooling reduces contention and reduces CPU utilization as it was found during profiling that creation of the XMLReader is expensive. Also, removed static methods on UIMASerializer. Each UIMA AS component that needs serialization/deserialization creates its owns instance of  UIMASerializer.

> Optimize UIMA AS XMI Cas serialization by reusing an instance of a XML Parser
> -----------------------------------------------------------------------------
>
>                 Key: UIMA-1200
>                 URL: https://issues.apache.org/jira/browse/UIMA-1200
>             Project: UIMA
>          Issue Type: Bug
>          Components: Async Scaleout
>            Reporter: Jerry Cwiklik
>         Attachments: uimaj-as-activemq-UIMA-1200-patch.txt, uimaj-as-core-UIMA-1200-patch.txt, uimaj-as-jms-UIMA-1200-patch.txt
>
>
> Current UIMA AS code creates an instance of XML parser for every CAS serialization/deserialization. Profiling revealed that creating a parser is CPU intensive. To avoid this create one or more instances of XML parser once and pool it for reuse. 

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.