[jira] [Commented] (PHOENIX-3834) Provide a way to collect stats and not use it for query parallelization

[Samarth Jain (JIRA)]

    [ 
https://issues.apache.org/jira/browse/PHOENIX-3834?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16004058#comment-16004058
 ] 

Samarth Jain commented on PHOENIX-3834:
---------------------------------------

With the help of this config, we could also possibly prevent loading of the 
stats cache. Today we issue an RPC for loading the stats cache even though we 
may not really be using the stats (by virtue of having a large guide post width 
~20GB)

{code}
protected class StatsLoader extends CacheLoader<GuidePostsKey, GuidePostsInfo> {
        @Override
        public GuidePostsInfo load(GuidePostsKey statsKey) throws Exception {
            @SuppressWarnings("deprecation")
            HTableInterface statsHTable = 
queryServices.getTable(SchemaUtil.getPhysicalName(
                    PhoenixDatabaseMetaData.SYSTEM_STATS_NAME_BYTES,
                            queryServices.getProps()).getName());
            try {
                GuidePostsInfo guidePostsInfo = 
StatisticsUtil.readStatistics(statsHTable, statsKey,
                        HConstants.LATEST_TIMESTAMP);
                traceStatsUpdate(statsKey, guidePostsInfo);
                return guidePostsInfo;
            }

{code}









> Provide a way to collect stats and not use it for query parallelization
> -----------------------------------------------------------------------
>
>                 Key: PHOENIX-3834
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-3834
>             Project: Phoenix
>          Issue Type: Improvement
>            Reporter: Samarth Jain
>            Assignee: Samarth Jain
>




--
This message was sent by Atlassian JIRA
(v6.3.15#6346)