[ 
https://issues.apache.org/jira/browse/HIVE-12631?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16034000#comment-16034000
 ] 

Teddy Choi commented on HIVE-12631:
-----------------------------------

Thank you for feedback, [~sershe].

1. I will merge OrcAcidColumnVectorProducer with OrcColumnVectorProducer.
2. DeleteEventsOverflowMemoryException is thrown when totalDeleteEventCount is 
larger than maxEventsInMemory, which is configurable by 
ConfVars.HIVE_TRANSACTIONAL_NUM_EVENTS_IN_MEMORY. So it will not cause OOM. It 
may be a better idea to return a boolean value to avoid confusion from the 
exception name.
3. The BitSet part is copied from VectorizedOrcAcidRowBatchReader of 
HIVE-14233. I will consider replacing both of them with boolean arrays.
4. I will make canUseLlapIo to return true for V2 base files and add V2 base 
file tests in llap_acid.q to ensure V2 support.

> LLAP: support ORC ACID tables
> -----------------------------
>
>                 Key: HIVE-12631
>                 URL: https://issues.apache.org/jira/browse/HIVE-12631
>             Project: Hive
>          Issue Type: Bug
>          Components: llap, Transactions
>            Reporter: Sergey Shelukhin
>            Assignee: Teddy Choi
>         Attachments: HIVE-12631.1.patch, HIVE-12631.2.patch, 
> HIVE-12631.3.patch, HIVE-12631.4.patch, HIVE-12631.5.patch, 
> HIVE-12631.6.patch, HIVE-12631.7.patch, HIVE-12631.8.patch, HIVE-12631.8.patch
>
>
> LLAP uses a completely separate read path in ORC to allow for caching and 
> parallelization of reads and processing. This path does not support ACID. As 
> far as I remember ACID logic is embedded inside ORC format; we need to 
> refactor it to be on top of some interface, if practical; or just port it to 
> LLAP read path.
> Another consideration is how the logic will work with cache. The cache is 
> currently low-level (CB-level in ORC), so we could just use it to read bases 
> and deltas (deltas should be cached with higher priority) and merge as usual. 
> We could also cache merged representation in future.



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

Reply via email to