You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@hive.apache.org by "KaiXu (JIRA)" <ji...@apache.org> on 2016/08/18 01:45:21 UTC

[jira] [Created] (HIVE-14567) After enabling Hive Parquet Vectorization, POWER_TEST of query24 in TPCx-BB(BigBench) failed with 1TB scale factor, but successful with 3TB scale factor

KaiXu created HIVE-14567:
----------------------------

             Summary: After enabling Hive Parquet Vectorization, POWER_TEST of query24 in TPCx-BB(BigBench) failed with 1TB scale factor, but successful with 3TB scale factor
                 Key: HIVE-14567
                 URL: https://issues.apache.org/jira/browse/HIVE-14567
             Project: Hive
          Issue Type: Bug
          Components: File Formats, Hive
    Affects Versions: 2.1.0
         Environment: Apache Hadoop2.6.0
Apache Hive2.1.0
JDK1.8.0_73
TPCx-BB 1.0.1
            Reporter: KaiXu
            Priority: Critical


We use TPCx-BB(BigBench) to evaluate the performance of Hive Parquet Vectorization in our local cluster(E5-2699 v3, 256G, 72 vcores, 1 master node + 5 worker nodes). During our performance test, we found that query24 in TPCx-BB failed with 1TB scale factor, but it is successful with 3TB scale factor on the same conditions. We retried with 100GB/10GB/1GB scale factor, they all failed. That is to say, with smaller data scale it fails but larger data scale successes, which seems very unusual.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)