[
https://issues.apache.org/jira/browse/FLINK-11856?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kurt Young updated FLINK-11856:
-------------------------------
Description: Once we have BinaryRow, we should also introduce the hash
table which stored all rows in binary format. It has multiple benefits if we
knows the binary format about the build side row and probe side row. For
example, we can directly compare on byte arrays when we want to check if build
key and probe key equals. We can also eliminate lots of deserialize cost when
we want to get a row instance from some binary data.
> Introduce BinaryHashTable to batch table runtime
> ------------------------------------------------
>
> Key: FLINK-11856
> URL: https://issues.apache.org/jira/browse/FLINK-11856
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Operators
> Reporter: Kurt Young
> Assignee: Kurt Young
> Priority: Major
> Labels: pull-request-available
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Once we have BinaryRow, we should also introduce the hash table which stored
> all rows in binary format. It has multiple benefits if we knows the binary
> format about the build side row and probe side row. For example, we can
> directly compare on byte arrays when we want to check if build key and probe
> key equals. We can also eliminate lots of deserialize cost when we want to
> get a row instance from some binary data.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
