Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-7757

Functional index pointing to tables

        Details

        • Type: Task
        • Status: Open
        • Priority: Major
        • Resolution: Unresolved
        • Fix Version/s: None
        • Component/s: None
        • Labels:
          None

          Description

          This feature enable to leverage storage engine features at the SQL Layer.

          Spatial index for any storage :

           ALTER TABLE t1_innodb ADD FUNCTIONAL INTERNAL INDEX idx1(c1) UNIQUE PIVOT(primary)  SPATIAL USING MyISAM

          Clustered index for any storage:

           ALTER TABLE t1_innodb ADD FUNCTIONAL INTERNAL INDEX idx1(c1) UNIQUE PIVOT(primary) CLUSTERED=YES USING TokuDB
          • This create a new table idx1 (PK,c1).
          • This table have storage engine defined with USING keyword
          • This table have a spatial index or a clustered index on c1
          • This table is auto updated when define as INTERNAL and is not updated with keyword EXTERNAL
          • This table is a 1 to 1 relation on the primary key of the indexed table keyword UNIQUE PIVOT(primary)

          Parser rewrite query and inject subquery for QP using t1_innodb and filters on column c1

          like :

           SELECT * FROM  t1_innodb where MBR_CONTAINS(GEOM_FROMTEXT(blabal) ,c1)

          rewrite to :

           SELECT * FROM  t1_innodb  JOIN (SELECT idx1.c1  FROM  idx1 WHERE GEOM_FROMTEXT(blabal) ,c1))  fct_index1  ON  fctindex1.pk=t1_innodb.pk  WHERE MBR_CONTAINS(GEOM_FROMTEXT(blabal) ,c1)

          Performance improvements using external table materialization 

           ALTER TABLE t1_innodb ADD FUNCTIONAL EXTERNAL INDEX idx1(t2.c2, t3.c3) NON UNIQUE PIVOT (fctT1.c1=t1_innodb.c1)
          • This index point to an already exiting table idx1 (c1,c2,c3).
          • This table is not updated by MariaDB but via triggers, scheduler, sphinx_se
          • This table have same columns name and value found in other tables EXTERNAL INDEX idx1(t2.c2, t3.c3)
          • This table is joinable on such columns PIVOT (idx1.c1=t1_innodb.c1)
          • This table is a 1 ton n relation defined with keyword NON UNIQUE
          • This table is having an index on c2,c3
           SELECT * FROM  t1_innodb JOIN t2 USING(id) JOIN t3 USING(id) WHERE t2.c1=4 AND t3.c2> 10

          Rewrite to

           SELECT* FROM  t1_innodb JOIN t2 USING(id) JOIN t3 USING(id) JOIN (SELECT DISTINCT t1.column  FROM  t1 WHERE WHERE t2.c1=4 AND t3.c2> 10 )   fct_index1  ON  fctindex1.c1=t1_innodb.c1 WHERE t2.c1=4 AND t3.c2> 10

            Gliffy Diagrams

              Attachments

                Activity

                There are no comments yet on this issue.

                  People

                  • Assignee:
                    Unassigned
                    Reporter:
                    stephane@skysql.com VAROQUI Stephane
                  • Votes:
                    0 Vote for this issue
                    Watchers:
                    2 Start watching this issue

                    Dates

                    • Created:
                      Updated: