[jira] [Commented] (ASTERIXDB-2310) Use Primary Key Index to Enforce Insert Key Uniqueness

["ASF subversion and git services (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ASTERIXDB-2310?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16823201#comment-16823201 ] 

ASF subversion and git services commented on ASTERIXDB-2310:
------------------------------------------------------------

Commit 89cec5899945cf41e3269b24c2ac138dc52b97f7 in asterixdb's branch refs/heads/master from luochen
[ https://gitbox.apache.org/repos/asf?p=asterixdb.git;h=89cec58 ]

[ASTERIXDB-2310][STO]Enforce Key Uniquness using PKIndex

- user model changes: no
- storage format changes: yes. Primary key index
now has bloom filters.
- interface changes: no

Details:
- Add bloom filters to primary key index.
- Introduce LSMPrimaryInsertOperator to separate uniqueness check from
the primary index. When the primary key index is available, it will be
used for uniqueness check. This implementation of this operation is
similar to LSMPrimaryUpsertOperator.

Change-Id: I7a52bb75ee5b14521972999df2f45ba62adc5af1
Reviewed-on: https://asterix-gerrit.ics.uci.edu/2453
Contrib: Jenkins <je...@fulliautomatix.ics.uci.edu>
Tested-by: Jenkins <je...@fulliautomatix.ics.uci.edu>
Integration-Tests: Jenkins <je...@fulliautomatix.ics.uci.edu>
Reviewed-by: Murtadha Hubail <mh...@apache.org>


> Use Primary Key Index to Enforce Insert Key Uniqueness
> ------------------------------------------------------
>
>                 Key: ASTERIXDB-2310
>                 URL: https://issues.apache.org/jira/browse/ASTERIXDB-2310
>             Project: Apache AsterixDB
>          Issue Type: Improvement
>          Components: ING - Ingestion, STO - Storage
>            Reporter: Chen Luo
>            Assignee: Chen Luo
>            Priority: Major
>              Labels: triaged
>
> **Currently when ingesting data using INSERT operations, we always check the primary index to ensure key uniqueness. However, this implies in most cases all ingested records might be accessed, which will slow down the ingestion performance a lot when the records cannot be cached. To handle this, we can enforce key uniqueness by checking the primary key index, which is much smaller and can be more easily cached.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)