[
https://issues.apache.org/jira/browse/HIVE-23042?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17061744#comment-17061744
]
Denys Kuzmenko commented on HIVE-23042:
---------------------------------------
[~pvary], why not use batching (at least for Postgres/Oracle)? Anyways you
already have an "if".
> Merge queries to a single one for updating MIN_OPEN_TXNS table
> --------------------------------------------------------------
>
> Key: HIVE-23042
> URL: https://issues.apache.org/jira/browse/HIVE-23042
> Project: Hive
> Issue Type: Bug
> Components: Transactions
> Reporter: Peter Vary
> Assignee: Peter Vary
> Priority: Major
> Attachments: HIVE-23042.patch
>
>
> When opening a new transaction we issue 2 queries to update the MIN_OPEN_TXN
> table.
> {code}
> <SELECT MIN("TXN_ID") FROM "TXNS" WHERE "TXN_STATE" = 'o'>
> <insert into "MIN_HISTORY_LEVEL" ("MHL_TXNID", "MHL_MIN_OPEN_TXNID")
> values(763, 763)>
> {code}
> This could be archived with a single query faster, if we do not open
> transactions in batch, like:
> {code}
> <INSERT INTO "MIN_HISTORY_LEVEL" ("MHL_TXNID", "MHL_MIN_OPEN_TXNID")
> SELECT ?, MIN("TXN_ID") FROM "TXNS" WHERE "TXN_STATE" = 'o'>
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
