🗓️ 31082025 1113
📎

sql_recursion

How Recursive CTEs Work

Recursive CTEs (WITH RECURSIVE) are commonly used for hierarchies and graphs.
Think of them like a controlled loop.

Step	What happens	Example in concept
Seed / Anchor	Start with a base set of rows.	SELECT id, parent_id FROM table_edges WHERE depth=0
Recursive step	Join previous results back to the table to get next-level rows.	JOIN table_edges e ON e.id = a.parent_id
Termination	Stops when no new rows are returned.	Implicit
Materialization	The engine stores all “visited” rows in a working table to avoid infinite loops.	Stored in memory or spilled to disk if big

Why Memory Usage Spikes

Recursive CTEs can be expensive because each iteration accumulates rows.
Memory pressure builds up due to these factors:

Cause	Why it’s heavy	Example scenario
Large recursion frontier	Each iteration multiplies rows.	Wide org chart, many ancestors.
Wide carried rows	More columns = larger working table.	Carrying name, description, group_name inside recursion unnecessarily.
Early ORDER BY	Sorting intermediate results per iteration is expensive.	Sorting inside the recursion instead of at the end.
Late filtering	Include irrelevant rows → filter later.	Join everything, then WHERE group_name IS NOT NULL.
No deduplication	Same (id, parent_id) processed multiple times.	No DISTINCT until very late.
Missing indexes	Each recursive step does a full table scan.	No index on id or parent_id.

---

References