EXPLAIN PLAN

EXPLAIN PLAN displays the plans used for SELECT statements, indexes, and materialized views.

Syntax

Note that the FOR keyword is required if the PLAN keyword is present. In other words, the following three statements are equivalent:

EXPLAIN <explainee>;
EXPLAIN PLAN FOR <explainee>;
EXPLAIN OPTIMIZED PLAN FOR <explainee>;

Explained object

The following object types can be explained.

Output format

You can select between JSON and TEXT for the output format of EXPLAIN PLAN. Non-text output is more machine-readable and can be parsed by common graph visualization libraries, while formatted text is more human-readable.

Explained stage

This stage determines the query optimization stage at which the plan snapshot will be taken.

Output modifiers

Output modifiers act as boolean toggles and can be combined in order to slightly tweak the information and rendering style of the generated explanation output.

Note that most modifiers are currently only supported for the AS TEXT output.

Details

To execute SELECT statements, Materialize generates a plan consisting of operators that interface with our underlying Differential dataflow engine. EXPLAIN PLAN lets you see the plan for a given query, which can provide insight into Materialize’s behavior for specific queries, e.g. performance.

Query compilation pipeline

The job of the Materialize planner is to turn SQL code into a differential dataflow program. We get there via a series of progressively lower-level plans:

SQL ⇒ raw plan ⇒ decorrelated plan ⇒ optimized plan ⇒ physical plan ⇒ dataflow

From SQL to raw plan

In this stage, the planner:

• Replaces SQL variable names with column numbers.
• Infers the type of each expression.
• Chooses the correct overload for each function.

From raw plan to decorrelated plan

In this stage, the planner:

• Replaces subqueries and lateral joins with non-nested operations.
• Replaces OUTER joins with lower-level operations.
• Replaces global aggregate default values with lower-level operations.

From decorrelated plan to optimized plan

In this stage, the planner performs various optimizing rewrites:

• Coalesces joins.
• Chooses join order and implementation.
• Fuses adjacent operations.
• Removes redundant operations.
• Pushes down predicates.
• Evaluates any operations on constants.

From optimized plan to physical plan

In this stage, the planner:

• Decides on the exact execution details of each operator, and maps plan operators to differential dataflow operators.
• Makes the final choices about creating or reusing arrangements.

From physical plan to dataflow

In the final stage, the planner:

• Renders an actual dataflow from the physical plan, and
• Installs the new dataflow into the running system.

The rendering step does not make any further optimization choices, as the physical plan is meant to be a definitive and complete description of the rendered dataflow.

Fast path queries

Queries are sometimes implemented using a fast path. In this mode, the program that implements the query will just hit an existing index, transform the results, and optionally apply a finishing action. For fast path queries, all of these actions happen outside of the regular dataflow engine. The fast path is indicated by an “Explained Query (fast path):” heading before the explained query in the EXPLAIN OPTIMIZED PLAN and EXPLAIN PHYSICAL PLAN result.

Explained Query (fast path):
  Project (#0, #1)
    ReadIndex on=materialize.public.t1 t1_x_idx=[lookup value=(5)]

Used Indexes:
  - materialize.public.t1_x_idx (lookup)

Reading decorrelated and optimized plans

Materialize plans are directed, potentially cyclic, graphs of operators. Each operator in the graph receives inputs from zero or more other operators and produces a single output. Sub-graphs where each output is consumed only once are rendered as tree-shaped fragments. Sub-graphs consumed more than once are represented as common table expressions (CTEs). In the example below, the CTE l0 represents a linear sub-plan (a chain of Get, Filter, and Project operators) which is used in both inputs of a self-join.

With
  cte l0 =
    Project (#0, #1)
      Filter (#0 > #2)
        ReadStorage materialize.public.t
Return
  Join on=(#1 = #2)
    Get l0
    Get l0

Note that CTEs in optimized plans do not directly correspond to CTEs in your original SQL query: For example, CTEs might disappear due to inlining (i.e., when a CTE is used only once, its definition is copied to that usage site); new CTEs can appear due to the optimizer recognizing that a part of the query appears more than once (aka common subexpression elimination). Also, certain SQL-level concepts, such as outer joins or subqueries, do not have an explicit representation in optimized plans, and are instead expressed as a pattern of operators involving CTEs. CTE names are always l0, l1, l2, …, and do not correspond to SQL-level CTE names.

Many operators need to refer to columns in their input. These are displayed like #3 for column number 3. (Columns are numbered starting from column 0). To get a better sense of columns assigned to Map operators, it might be useful to request the arity output modifier.

Each operator can also be annotated with additional metadata. Details are shown by default in the EXPLAIN PHYSICAL PLAN output, but are hidden elsewhere. In EXPLAIN OPTIMIZED PLAN, details about the implementation in the Join operator can be requested with the join implementations output modifier (that is, EXPLAIN OPTIMIZED PLAN WITH (join implementations) FOR ...).

Join on=(#1 = #2 AND #3 = #4) type=delta
  implementation
    %0:t » %1:u[#0]K » %2:v[#0]K
    %1:u » %0:t[#1]K » %2:v[#0]K
    %2:v » %1:u[#1]K » %0:t[#1]K
  ArrangeBy keys=[[#1]]
    ReadStorage materialize.public.t
  ArrangeBy keys=[[#0], [#1]]
    ReadStorage materialize.public.u
  ArrangeBy keys=[[#0]]
    ReadStorage materialize.public.v

The %0, %1, etc. refer to each of the join inputs. A differential join shows one join path, which is simply a sequence of binary joins (each of whose results need to be maintained as state). A delta join shows a join path for each of the inputs. The expressions in a bracket show the key for joining with that input. The letters after the brackets indicate the input characteristics used for join ordering. U means unique, the number of Ks means the key length, A means already arranged (e.g., an index exists). The small letters refer to filter characteristics: equality to a literal, like, is null, inequality to a literal, any filter.

A plan can optionally end with a finishing action, which can sort, limit and project the result data. This operator is special, as it can only occur at the top of the plan. Finishing actions are executed outside the parallel dataflow that implements the rest of the plan.

Finish order_by=[#1 asc nulls_last, #0 desc nulls_first] limit=5 output=[#0, #1]
  CrossJoin
    ReadStorage materialize.public.r
    ReadStorage materialize.public.s

Below the plan, a “Used indexes” section indicates which indexes will be used by the query, and in what way.

Reference: Plan operators

Materialize offers several output formats for EXPLAIN and debugging. LIR plans as rendered in mz_introspection.mz_lir_mapping are deliberately succinct, while the plans in other formats give more detail.

The decorrelated and optimized plans from EXPLAIN DECORRELATED PLAN FOR ..., EXPLAIN LOCALLY OPTIMIZED PLAN FOR ..., and EXPLAIN OPTIMIZED PLAN FOR ... are in a mid-level representation that is closer to LIR than SQL. The raw plans from EXPLAIN RAW PLAN FOR ... are closer to SQL (and therefore less indicative of how the query will actually run).

Constant
- ((1, 2) x 2)
- (3, 4)

Constant 2 rows

Constant
- ((1, 2) x 2)
- (3, 4)

Examples

For the following examples, let’s assume that you have the auction house load generator created in your current environment.

Explaining a SELECT query

Let’s start with a simple join query that lists the total amounts bid per buyer.

Explain the optimized plan as text:

EXPLAIN
SELECT a.id, sum(b.amount) FROM accounts a JOIN bids b ON(a.id = b.buyer) GROUP BY a.id;

Same as above, but a bit more verbose:

EXPLAIN PLAN
SELECT a.id, sum(b.amount) FROM accounts a JOIN bids b ON(a.id = b.buyer) GROUP BY a.id;

Same as above, but even more verbose:

EXPLAIN OPTIMIZED PLAN AS TEXT FOR
SELECT a.id, sum(b.amount) FROM accounts a JOIN bids b ON(a.id = b.buyer) GROUP BY a.id;

Same as above, but every sub-plan is annotated with its schema types:

EXPLAIN WITH(types) FOR
SELECT a.id, sum(b.amount) FROM accounts a JOIN bids b ON(a.id = b.buyer) GROUP BY a.id;

Explain the physical plan as text:

EXPLAIN PHYSICAL PLAN FOR
SELECT a.id, sum(b.amount) FROM accounts a JOIN bids b ON(a.id = b.buyer) GROUP BY a.id;

Explaining an index on a view

Let’s create a view with an index for the above query.

-- create the view
CREATE VIEW my_view AS
SELECT a.id, sum(b.amount) FROM accounts a JOIN bids b ON(a.id = b.buyer) GROUP BY a.id;
-- create an index on the view
CREATE INDEX my_view_idx ON my_view(id);

You can inspect the plan of the dataflow that will maintain your index with the following statements.

Explain the optimized plan as text:

EXPLAIN
INDEX my_view_idx;

Same as above, but a bit more verbose:

EXPLAIN PLAN FOR
INDEX my_view_idx;

Same as above, but even more verbose:

EXPLAIN OPTIMIZED PLAN AS TEXT FOR
INDEX my_view_idx;

Same as above, but every sub-plan is annotated with its schema types:

EXPLAIN WITH(types) FOR
INDEX my_view_idx;

Explain the physical plan as text:

EXPLAIN PHYSICAL PLAN FOR
INDEX my_view_idx;

Explaining a materialized view

Let’s create a materialized view for the above SELECT query.

CREATE MATERIALIZED VIEW my_mat_view AS
SELECT a.id, sum(b.amount) FROM accounts a JOIN bids b ON(a.id = b.buyer) GROUP BY a.id;

You can inspect the plan of the dataflow that will maintain your view with the following statements.

Explain the optimized plan as text:

EXPLAIN
MATERIALIZED VIEW my_mat_view;

Same as above, but a bit more verbose:

EXPLAIN PLAN FOR
MATERIALIZED VIEW my_mat_view;

Same as above, but even more verbose:

EXPLAIN OPTIMIZED PLAN AS TEXT FOR
MATERIALIZED VIEW my_mat_view;

Same as above, but every sub-plan is annotated with its schema types:

EXPLAIN WITH(types)
MATERIALIZED VIEW my_mat_view;

Explain the physical plan as text:

EXPLAIN PHYSICAL PLAN FOR
MATERIALIZED VIEW my_mat_view;

Privileges

The privileges required to execute this statement are:

• USAGE privileges on the schemas that all relations in the explainee are contained in.