Lindorm:Overview of query plans

Logical plan

The logical plan shows the query structure after rewriting, before optimization. Lindorm SQL generates it from the abstract syntax tree and may apply rewrites such as:

• Replacing a constant expression with its computed value

• Replacing AVG with equivalent SUM and COUNT expressions

Use EXPLAIN WITHOUT IMPLEMENTATION FOR to view the logical plan:

EXPLAIN WITHOUT IMPLEMENTATION FOR
  SELECT LOCATE('9378', p1) AS lp, COUNT(c1)
  FROM (SELECT * FROM test WHERE p1 LIKE '3_%')
  GROUP BY lp
  ORDER BY lp;

Output:

********************* 1. row *********************
PLAN: LogicalSort(sort0=[$0], dir0=[ASC])
  LogicalAggregate(group=[{0}], EXPR$1=[COUNT($1)])
    LogicalProject(lp=[LOCATE('9378', $0)], c1=[$1])
      LogicalFilter(condition=[LIKE($0, '3_%')])
        LindormTableScan(table=[[db, test]])

For the full EXPLAIN syntax, see EXPLAIN.

Physical plan

The physical plan is the optimized, executable version of the query. The optimizer selects the best execution mode based on rule-based or cost-based policies. This is the default output of EXPLAIN and shows the complete execution path: which operators run in Lindorm SQL, which operations are pushed down to LindormTable or LindormTSDB, the index chosen, the scan range, and the filter conditions.

Use EXPLAIN FOR (or just EXPLAIN) to view the physical plan:

EXPLAIN
  SELECT LOCATE('9378', p1) AS lp, COUNT(c1)
  FROM (SELECT * FROM test WHERE p1 LIKE '3_%')
  GROUP BY lp
  ORDER BY lp;

Output:

********************* 1. row *********************
PLAN: EnumerableSort(sort0=[$0], dir0=[ASC])
  EnumerableAggregate(group=[{0}], EXPR$1=[COUNT($1)])
    EnumerableCalc(expr#0..3=[{inputs}], expr#4=['9378'], expr#5=[LOCATE($t4, $t0)], lp=[$t5], c1=[$t1])
      EnumerableLindormDirectQuery(
explanation=[SELECT p1,c1,c2,c3 from test where ((p1 < 4) AND (p1 >= 3) AND (p1 LIKE 3_%)) supportEmptyResult true
Candidate tables:
    Data table: test, scores=1..0..1, need query back=false, sort type=FORWARD
Chose data table [test].
---
SELECT p1,c1,c2,c3 from test supportEmptyResult true
SingleScan on test
    ranges: [3\x00, 4\x00]
    filter: ((p1 < 4) AND (p1 >= 3) AND (p1 LIKE 3_%))
])

Type plan

The type plan contains only the output column types for each column in the result set. It is generated based on query types and is applicable in specific scenarios. It is the most lightweight of the three depths and is useful for quickly verifying the output schema without inspecting the full execution tree.

Use EXPLAIN WITH TYPE FOR to view the type plan:

EXPLAIN WITH TYPE FOR
  SELECT LOCATE('9378', p1) AS lp, COUNT(c1)
  FROM (SELECT * FROM test WHERE p1 LIKE '3_%')
  GROUP BY lp
  ORDER BY lp;

Output:

********************* 1. row *********************
PLAN: lp INTEGER NOT NULL,
EXPR$1 BIGINT NOT NULL

For the full EXPLAIN syntax, see EXPLAIN.

General-purpose operators

These operators handle computing operations that run in Lindorm SQL — sorting, aggregation, projection, and calculation.

Extended operators