[IGNITE-12620] Calcite integration. Index Nested Loop Join/Hash Join

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Ignite Flags: Docs Required, Release Notes Required
• Assignee: Igor Seliverstov
• Reporter: Igor Seliverstov
• Reviewer: Alexey Goncharuk
• Created: 04/Feb/20 13:57
• Updated: 30/Nov/20 15:14
• Resolved: 03/Sep/20 12:30

Description

We may implement the feature the next way:

1. For each row from left node consume whole dataset from right node
2. Pass join condition as an argument of request() method of the right node
3. In case a data source at right is an index scan
   1. If there is no cursor opened - create a new cursor using bounds from request 
   2. If there is an existing cursor - just check the cursor was opened using the same condition as passed one.
   3. After the right node signals EOD consume a next row from left and repeat from p 2.
4. In case a data source at right is a table scan with huge amount of rows
   1. If there is no cursor opened - create a new cursor ignoring passed condition
   2. If there is an existing cursor - just check the cursor was opened using the same condition as passed one.
   3. After the right node signals EOD consume a next row from left and repeat from p 2.
5. In case a data source at right is remote / is a table scan with small number of rows/ is a filter node with good enough selectivity
   1. Create a hash index for a data source
   2. If there is no cursor opened - create a new cursor using bounds from request 
   3. If there is an existing cursor - just check the cursor was opened using the same condition as passed one.
   4. After the right node signals EOD consume a next row from left and repeat from p 2.

 Consider implementation specifics at optimization time, choose the cheapest variant

Issue Links

• is part of: IGNITE-12248 Apache Calcite based query execution engine
• links to: GitHub Pull Request #8210