[
https://issues.apache.org/jira/browse/ASTERIXDB-2924?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17391785#comment-17391785
]
ASF subversion and git services commented on ASTERIXDB-2924:
------------------------------------------------------------
Commit dd99b2445416be08591713563787cc8dedebc8f5 in asterixdb's branch
refs/heads/master from ggalvizo
[ https://gitbox.apache.org/repos/asf?p=asterixdb.git;h=dd99b24 ]
[ASTERIXDB-2924][IDX] Multiple INLJs are now recognized.
- user mode changes: no
- storage format changes: no
- interface changes: no
WLOG, two datasets that can be optimized with an INLJ using the same
probe are now recognized and transformed into a double INLJ plan.
Change-Id: I1da6174b07963ec611677d7906c7304c46a2a382
Reviewed-on: https://asterix-gerrit.ics.uci.edu/c/asterixdb/+/12424
Tested-by: Jenkins <[email protected]>
Integration-Tests: Jenkins <[email protected]>
Reviewed-by: Dmitry Lychagin <[email protected]>
> Ternary joins with same probe using INLJ
> ----------------------------------------
>
> Key: ASTERIXDB-2924
> URL: https://issues.apache.org/jira/browse/ASTERIXDB-2924
> Project: Apache AsterixDB
> Issue Type: Improvement
> Components: IDX - Indexes
> Reporter: Glenn Justo Galvizo
> Assignee: Glenn Justo Galvizo
> Priority: Major
> Fix For: 0.9.7
>
>
> Given the DDLs:
> {code:java}
> DROP DATAVERSE TestDataverse IF EXISTS;
> CREATE DATAVERSE TestDataverse;
> USE TestDataverse;
> CREATE TYPE GenericType AS { _id: bigint, c : bigint };
> CREATE DATASET IndexDatasetA (GenericType)
> PRIMARY KEY _id;
> CREATE DATASET IndexDatasetB (GenericType)
> PRIMARY KEY _id;
> CREATE DATASET ProbeDataset (GenericType)
> PRIMARY KEY _id;
> CREATE INDEX indexA
> ON IndexDatasetA (k : int);
> CREATE INDEX indexB
> ON IndexDatasetB (k : int);{code}
> The following ternary join query produces an INLJ plan for both indexed
> datasets. The join field from the probe is the probe dataset's primary key.
> {code:java}
> -- Query 1, ternary join w/ primary key on probe.
> FROM ProbeDataset P,
> IndexDatasetA A,
> IndexDatasetB B
> WHERE P._id /* +indexnl */ = A.k AND
> P._id /* +indexnl */ = B.k
> SELECT COUNT(*);{code}
> The following ternary join query produces an INLJ plan for only the left-most
> indexed dataset. The join field from the probe is a closed field from the
> probe dataset.
> {code:java}
> -- Query 2, ternary join w/ closed field on probe.
> FROM ProbeDataset P,
> IndexDatasetA A,
> IndexDatasetB B
> WHERE P.c /* +indexnl */ = A.k AND
> P.c /* +indexnl */ = B.k
> SELECT COUNT(*);
> {code}
> Both queries should produce two INLJs.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
