[SPARK-48047] Reduce memory pressure of TreeNode `tags` variable

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 4.0.0
• Fix Version/s: 4.0.0
• Component/s: Optimizer
• Labels: pull-request-available
• Assignee: Nick Young
• Reporter: Nick Young
• Created: 29/Apr/24 18:14
• Updated: 06/May/24 15:51
• Resolved: 01/May/24 05:07

Description

By default, the TreeNode `tags` variable is initialized to a `mutable.Map.empty`, which defaults to a map with an internal 16 element hashtable array. For plans with a high number of TreeNodes with empty tags, this adds unnecessary driver memory pressure, especially for large expression trees. Reduce the memory pressure by lazily initializing this variable.

Issue Links

• links to: GitHub Pull Request #46285