[
https://issues.apache.org/jira/browse/HIVE-4486?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13654668#comment-13654668
]
Gopal V commented on HIVE-4486:
-------------------------------
Closed that (wrong diff), opened as https://reviews.apache.org/r/11048/diff/
instead.
> FetchOperator slows down SMB map joins by 50% when there are many partitions
> ----------------------------------------------------------------------------
>
> Key: HIVE-4486
> URL: https://issues.apache.org/jira/browse/HIVE-4486
> Project: Hive
> Issue Type: Bug
> Components: Query Processor
> Affects Versions: 0.12.0
> Environment: Ubuntu LXC 12.10
> Reporter: Gopal V
> Priority: Minor
> Attachments: HIVE-4486.patch, smb-profile.html
>
>
> While looking at log files for SMB joins in hive, it was noticed that the
> actual join op didn't show up as a significant fraction of the time spent.
> Most of the time was spent parsing configuration files.
> To confirm, I put log lines in the HiveConf constructor and eventually made
> the following edit to the code
> {code}
> --- ql/src/java/org/apache/hadoop/hive/ql/exec/FetchOperator.java
> +++ ql/src/java/org/apache/hadoop/hive/ql/exec/FetchOperator.java
> @@ -648,8 +648,7 @@ public ObjectInspector getOutputObjectInspector() throws
> HiveException {
> * @return list of file status entries
> */
> private FileStatus[] listStatusUnderPath(FileSystem fs, Path p) throws
> IOException {
> - HiveConf hiveConf = new HiveConf(job, FetchOperator.class);
> - boolean recursive =
> hiveConf.getBoolVar(HiveConf.ConfVars.HADOOPMAPREDINPUTDIRRECURSIVE);
> + boolean recursive = false;
> if (!recursive) {
> return fs.listStatus(p);
> }
> {code}
> And re-ran my query to compare timings.
> || ||Before||After||
> |Cumulative CPU| 731.07 sec|386.0 sec|
> |Total time | 347.66 seconds | 218.855 seconds |
> |
> The query used was
> {code}INSERT OVERWRITE LOCAL DIRECTORY
> '/grid/0/smb/'
> select inv_item_sk
> from
> inventory inv
> join store_sales ss on (ss.ss_item_sk = inv.inv_item_sk)
> limit 100000
> ;
> {code}
> On a scale=2 tpcds data-set, where both store_sales & inventory are bucketed
> into 4 buckets, with store_sales split into 7 partitions and inventory into
> 261 partitions.
> 78% of all CPU time was spent within new HiveConf(). The yourkit profiler
> runs are attached.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
