<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>15.3. Parallel Plans</title><link rel="stylesheet" type="text/css" href="stylesheet.css" /><link rev="made" href="pgsql-docs@lists.postgresql.org" /><meta name="generator" content="DocBook XSL Stylesheets Vsnapshot" /><link rel="prev" href="when-can-parallel-query-be-used.html" title="15.2. When Can Parallel Query Be Used?" /><link rel="next" href="parallel-safety.html" title="15.4. Parallel Safety" /></head><body><div xmlns="http://www.w3.org/TR/xhtml1/transitional" class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center">15.3. Parallel Plans</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="when-can-parallel-query-be-used.html" title="15.2. When Can Parallel Query Be Used?">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="parallel-query.html" title="Chapter 15. Parallel Query">Up</a></td><th width="60%" align="center">Chapter 15. Parallel Query</th><td width="10%" align="right"><a accesskey="h" href="index.html" title="PostgreSQL 11.5 Documentation">Home</a></td><td width="10%" align="right"> <a accesskey="n" href="parallel-safety.html" title="15.4. Parallel Safety">Next</a></td></tr></table><hr></hr></div><div class="sect1" id="PARALLEL-PLANS"><div class="titlepage"><div><div><h2 class="title" style="clear: both">15.3. Parallel Plans</h2></div></div></div><div class="toc"><dl class="toc"><dt><span class="sect2"><a href="parallel-plans.html#PARALLEL-SCANS">15.3.1. Parallel Scans</a></span></dt><dt><span class="sect2"><a href="parallel-plans.html#PARALLEL-JOINS">15.3.2. Parallel Joins</a></span></dt><dt><span class="sect2"><a href="parallel-plans.html#PARALLEL-AGGREGATION">15.3.3. Parallel Aggregation</a></span></dt><dt><span class="sect2"><a href="parallel-plans.html#PARALLEL-APPEND">15.3.4. Parallel Append</a></span></dt><dt><span class="sect2"><a href="parallel-plans.html#PARALLEL-PLAN-TIPS">15.3.5. Parallel Plan Tips</a></span></dt></dl></div><p>
Because each worker executes the parallel portion of the plan to
completion, it is not possible to simply take an ordinary query plan
and run it using multiple workers. Each worker would produce a full
copy of the output result set, so the query would not run any faster
than normal but would produce incorrect results. Instead, the parallel
portion of the plan must be what is known internally to the query
optimizer as a <em class="firstterm">partial plan</em>; that is, it must be constructed
so that each process which executes the plan will generate only a
subset of the output rows in such a way that each required output row
is guaranteed to be generated by exactly one of the cooperating processes.
Generally, this means that the scan on the driving table of the query
must be a parallel-aware scan.
</p><div class="sect2" id="PARALLEL-SCANS"><div class="titlepage"><div><div><h3 class="title">15.3.1. Parallel Scans</h3></div></div></div><p>
The following types of parallel-aware table scans are currently supported.
</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
In a <span class="emphasis"><em>parallel sequential scan</em></span>, the table's blocks will
be divided among the cooperating processes. Blocks are handed out one
at a time, so that access to the table remains sequential.
</p></li><li class="listitem"><p>
In a <span class="emphasis"><em>parallel bitmap heap scan</em></span>, one process is chosen
as the leader. That process performs a scan of one or more indexes
and builds a bitmap indicating which table blocks need to be visited.
These blocks are then divided among the cooperating processes as in
a parallel sequential scan. In other words, the heap scan is performed
in parallel, but the underlying index scan is not.
</p></li><li class="listitem"><p>
In a <span class="emphasis"><em>parallel index scan</em></span> or <span class="emphasis"><em>parallel index-only
scan</em></span>, the cooperating processes take turns reading data from the
index. Currently, parallel index scans are supported only for
btree indexes. Each process will claim a single index block and will
scan and return all tuples referenced by that block; other process can
at the same time be returning tuples from a different index block.
The results of a parallel btree scan are returned in sorted order
within each worker process.
</p></li></ul></div><p>
Other scan types, such as scans of non-btree indexes, may support
parallel scans in the future.
</p></div><div class="sect2" id="PARALLEL-JOINS"><div class="titlepage"><div><div><h3 class="title">15.3.2. Parallel Joins</h3></div></div></div><p>
Just as in a non-parallel plan, the driving table may be joined to one or
more other tables using a nested loop, hash join, or merge join. The
inner side of the join may be any kind of non-parallel plan that is
otherwise supported by the planner provided that it is safe to run within
a parallel worker. Depending on the join type, the inner side may also be
a parallel plan.
</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
In a <span class="emphasis"><em>nested loop join</em></span>, the inner side is always
non-parallel. Although it is executed in full, this is efficient if
the inner side is an index scan, because the outer tuples and thus
the loops that look up values in the index are divided over the
cooperating processes.
</p></li><li class="listitem"><p>
In a <span class="emphasis"><em>merge join</em></span>, the inner side is always
a non-parallel plan and therefore executed in full. This may be
inefficient, especially if a sort must be performed, because the work
and resulting data are duplicated in every cooperating process.
</p></li><li class="listitem"><p>
In a <span class="emphasis"><em>hash join</em></span> (without the "parallel" prefix),
the inner side is executed in full by every cooperating process
to build identical copies of the hash table. This may be inefficient
if the hash table is large or the plan is expensive. In a
<span class="emphasis"><em>parallel hash join</em></span>, the inner side is a
<span class="emphasis"><em>parallel hash</em></span> that divides the work of building
a shared hash table over the cooperating processes.
</p></li></ul></div></div><div class="sect2" id="PARALLEL-AGGREGATION"><div class="titlepage"><div><div><h3 class="title">15.3.3. Parallel Aggregation</h3></div></div></div><p>
<span class="productname">PostgreSQL</span> supports parallel aggregation by aggregating in
two stages. First, each process participating in the parallel portion of
the query performs an aggregation step, producing a partial result for
each group of which that process is aware. This is reflected in the plan
as a <code class="literal">Partial Aggregate</code> node. Second, the partial results are
transferred to the leader via <code class="literal">Gather</code> or <code class="literal">Gather
Merge</code>. Finally, the leader re-aggregates the results across all
workers in order to produce the final result. This is reflected in the
plan as a <code class="literal">Finalize Aggregate</code> node.
</p><p>
Because the <code class="literal">Finalize Aggregate</code> node runs on the leader
process, queries which produce a relatively large number of groups in
comparison to the number of input rows will appear less favorable to the
query planner. For example, in the worst-case scenario the number of
groups seen by the <code class="literal">Finalize Aggregate</code> node could be as many as
the number of input rows which were seen by all worker processes in the
<code class="literal">Partial Aggregate</code> stage. For such cases, there is clearly
going to be no performance benefit to using parallel aggregation. The
query planner takes this into account during the planning process and is
unlikely to choose parallel aggregate in this scenario.
</p><p>
Parallel aggregation is not supported in all situations. Each aggregate
must be <a class="link" href="parallel-safety.html" title="15.4. Parallel Safety">safe</a> for parallelism and must
have a combine function. If the aggregate has a transition state of type
<code class="literal">internal</code>, it must have serialization and deserialization
functions. See <a class="xref" href="sql-createaggregate.html" title="CREATE AGGREGATE"><span class="refentrytitle">CREATE AGGREGATE</span></a> for more details.
Parallel aggregation is not supported if any aggregate function call
contains <code class="literal">DISTINCT</code> or <code class="literal">ORDER BY</code> clause and is also
not supported for ordered set aggregates or when the query involves
<code class="literal">GROUPING SETS</code>. It can only be used when all joins involved in
the query are also part of the parallel portion of the plan.
</p></div><div class="sect2" id="PARALLEL-APPEND"><div class="titlepage"><div><div><h3 class="title">15.3.4. Parallel Append</h3></div></div></div><p>
Whenever <span class="productname">PostgreSQL</span> needs to combine rows
from multiple sources into a single result set, it uses an
<code class="literal">Append</code> or <code class="literal">MergeAppend</code> plan node.
This commonly happens when implementing <code class="literal">UNION ALL</code> or
when scanning a partitioned table. Such nodes can be used in parallel
plans just as they can in any other plan. However, in a parallel plan,
the planner may instead use a <code class="literal">Parallel Append</code> node.
</p><p>
When an <code class="literal">Append</code> node is used in a parallel plan, each
process will execute the child plans in the order in which they appear,
so that all participating processes cooperate to execute the first child
plan until it is complete and then move to the second plan at around the
same time. When a <code class="literal">Parallel Append</code> is used instead, the
executor will instead spread out the participating processes as evenly as
possible across its child plans, so that multiple child plans are executed
simultaneously. This avoids contention, and also avoids paying the startup
cost of a child plan in those processes that never execute it.
</p><p>
Also, unlike a regular <code class="literal">Append</code> node, which can only have
partial children when used within a parallel plan, a <code class="literal">Parallel
Append</code> node can have both partial and non-partial child plans.
Non-partial children will be scanned by only a single process, since
scanning them more than once would produce duplicate results. Plans that
involve appending multiple results sets can therefore achieve
coarse-grained parallelism even when efficient partial plans are not
available. For example, consider a query against a partitioned table
which can be only be implemented efficiently by using an index that does
not support parallel scans. The planner might choose a <code class="literal">Parallel
Append</code> of regular <code class="literal">Index Scan</code> plans; each
individual index scan would have to be executed to completion by a single
process, but different scans could be performed at the same time by
different processes.
</p><p>
<a class="xref" href="runtime-config-query.html#GUC-ENABLE-PARALLEL-APPEND">enable_parallel_append</a> can be used to disable
this feature.
</p></div><div class="sect2" id="PARALLEL-PLAN-TIPS"><div class="titlepage"><div><div><h3 class="title">15.3.5. Parallel Plan Tips</h3></div></div></div><p>
If a query that is expected to do so does not produce a parallel plan,
you can try reducing <a class="xref" href="runtime-config-query.html#GUC-PARALLEL-SETUP-COST">parallel_setup_cost</a> or
<a class="xref" href="runtime-config-query.html#GUC-PARALLEL-TUPLE-COST">parallel_tuple_cost</a>. Of course, this plan may turn
out to be slower than the serial plan which the planner preferred, but
this will not always be the case. If you don't get a parallel
plan even with very small values of these settings (e.g. after setting
them both to zero), there may be some reason why the query planner is
unable to generate a parallel plan for your query. See
<a class="xref" href="when-can-parallel-query-be-used.html" title="15.2. When Can Parallel Query Be Used?">Section 15.2</a> and
<a class="xref" href="parallel-safety.html" title="15.4. Parallel Safety">Section 15.4</a> for information on why this may be
the case.
</p><p>
When executing a parallel plan, you can use <code class="literal">EXPLAIN (ANALYZE,
VERBOSE)</code> to display per-worker statistics for each plan node.
This may be useful in determining whether the work is being evenly
distributed between all plan nodes and more generally in understanding the
performance characteristics of the plan.
</p></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="when-can-parallel-query-be-used.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="parallel-query.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="parallel-safety.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">15.2. When Can Parallel Query Be Used? </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> 15.4. Parallel Safety</td></tr></table></div></body></html>
