[
https://issues.apache.org/jira/browse/HIVE-6980?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16452770#comment-16452770
]
Peter Vary commented on HIVE-6980:
----------------------------------
Performance test results using [~akolb]'s testing tool
([https://github.com/akolb1/hclient)|https://github.com/akolb1/hclient):],
dropping table with 10/100/1000 partitions
Before:
{code:java}
[main] INFO com.akolb.HMSBenchmark - using args -P 9084 -H localhost -d
test_db7 -L 10 -N 10 -W 1 -S dropTableWithPartitions.10
Operation AMean Mean Med Min Max
Err%
dropTableWithPartitions.10 0.000 119.8 117.3 106.3 143.3
8.807
[main] INFO com.akolb.HMSBenchmark - using args -P 9084 -H localhost -d
test_db7 -L 10 -N 100 -W 1 -S dropTableWithPartitions.100
Operation AMean Mean Med Min Max
Err%
dropTableWithPartitions.100 0.000 661.1 654.7 623.1 728.2
5.535
[main] INFO com.akolb.HMSBenchmark - using args -P 9084 -H localhost -d
test_db7 -L 10 -N 1000 -W 1 -S dropTableWithPartitions.1000
Operation AMean Mean Med Min Max
Err%
dropTableWithPartitions.1000 0.000 6256 6282 5969 6492
2.714 {code}
After:
{code:java}
[main] INFO com.akolb.HMSBenchmark - using args -P 9084 -H localhost -d
test_db7 -L 10 -N 10 -W 1 -S dropTableWithPartitions.10
Operation AMean Mean Med Min Max
Err%
dropTableWithPartitions.10 0.000 55.60 54.43 50.93 68.56
9.390
[main] INFO com.akolb.HMSBenchmark - using args -P 9084 -H localhost -d
test_db7 -L 10 -N 100 -W 1 -S dropTableWithPartitions.100
Operation AMean Mean Med Min Max
Err%
dropTableWithPartitions.100 0.000 70.30 69.46 67.85 75.43
3.426
[main] INFO com.akolb.HMSBenchmark - using args -P 9084 -H localhost -d
test_db7 -L 10 -N 1000 -W 1 -S dropTableWithPartitions.1000
Operation AMean Mean Med Min Max
Err%
dropTableWithPartitions.1000 0.000 267.6 267.7 259.5 272.9
1.652 {code}
> Drop table by using direct sql
> ------------------------------
>
> Key: HIVE-6980
> URL: https://issues.apache.org/jira/browse/HIVE-6980
> Project: Hive
> Issue Type: Improvement
> Components: Metastore
> Affects Versions: 0.12.0
> Reporter: Selina Zhang
> Assignee: Peter Vary
> Priority: Major
> Attachments: HIVE-6980.patch
>
>
> Dropping table which has lots of partitions is slow. Even after applying the
> patch of HIVE-6265, the drop table still takes hours (100K+ partitions).
> The fixes come with two parts:
> 1. use directSQL to query the partitions protect mode;
> the current implementation needs to transfer the Partition object to client
> and check the protect mode for each partition. I'd like to move this part of
> logic to metastore. The check will be done by direct sql (if direct sql is
> disabled, execute the same logic in the ObjectStore);
> 2. use directSQL to drop partitions for table;
> there maybe two solutions here:
> 1. add "DELETE CASCADE" in the schema. In this way we only need to delete
> entries from partitions table use direct sql. May need to change
> datanucleus.deletionPolicy = DataNucleus.
> 2. clean up the dependent tables by issue DELETE statement. This also needs
> to turn on datanucleus.query.sql.allowAll
> Both of above solutions should be able to fix the problem. The DELETE CASCADE
> has to change schemas and prepare upgrade scripts. The second solutions added
> maintenance cost if new tables added in the future releases.
> Please advice.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
