codekingpro/portable-devtools
115k
1<?xml version="1.0" encoding="UTF-8" standalone="no"?>2<!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 id="docContent" class="container-fluid col-10"><div 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 16.3 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 /></div><div class="sect1" id="PARALLEL-PLANS"><div class="titlepage"><div><div><h2 class="title" style="clear: both">15.3. Parallel Plans <a href="#PARALLEL-PLANS" class="id_link">#</a></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>3 Because each worker executes the parallel portion of the plan to4 completion, it is not possible to simply take an ordinary query plan5 and run it using multiple workers. Each worker would produce a full6 copy of the output result set, so the query would not run any faster7 than normal but would produce incorrect results. Instead, the parallel8 portion of the plan must be what is known internally to the query9 optimizer as a <em class="firstterm">partial plan</em>; that is, it must be constructed10 so that each process that executes the plan will generate only a11 subset of the output rows in such a way that each required output row12 is guaranteed to be generated by exactly one of the cooperating processes.13 Generally, this means that the scan on the driving table of the query14 must be a parallel-aware scan.15 </p><div class="sect2" id="PARALLEL-SCANS"><div class="titlepage"><div><div><h3 class="title">15.3.1. Parallel Scans <a href="#PARALLEL-SCANS" class="id_link">#</a></h3></div></div></div><p>16 The following types of parallel-aware table scans are currently supported.17 18 </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>19 In a <span class="emphasis"><em>parallel sequential scan</em></span>, the table's blocks will20 be divided into ranges and shared among the cooperating processes. Each21 worker process will complete the scanning of its given range of blocks before22 requesting an additional range of blocks.23 </p></li><li class="listitem"><p>24 In a <span class="emphasis"><em>parallel bitmap heap scan</em></span>, one process is chosen25 as the leader. That process performs a scan of one or more indexes26 and builds a bitmap indicating which table blocks need to be visited.27 These blocks are then divided among the cooperating processes as in28 a parallel sequential scan. In other words, the heap scan is performed29 in parallel, but the underlying index scan is not.30 </p></li><li class="listitem"><p>31 In a <span class="emphasis"><em>parallel index scan</em></span> or <span class="emphasis"><em>parallel index-only32 scan</em></span>, the cooperating processes take turns reading data from the33 index. Currently, parallel index scans are supported only for34 btree indexes. Each process will claim a single index block and will35 scan and return all tuples referenced by that block; other processes can36 at the same time be returning tuples from a different index block.37 The results of a parallel btree scan are returned in sorted order38 within each worker process.39 </p></li></ul></div><p>40 41 Other scan types, such as scans of non-btree indexes, may support42 parallel scans in the future.43 </p></div><div class="sect2" id="PARALLEL-JOINS"><div class="titlepage"><div><div><h3 class="title">15.3.2. Parallel Joins <a href="#PARALLEL-JOINS" class="id_link">#</a></h3></div></div></div><p>44 Just as in a non-parallel plan, the driving table may be joined to one or45 more other tables using a nested loop, hash join, or merge join. The46 inner side of the join may be any kind of non-parallel plan that is47 otherwise supported by the planner provided that it is safe to run within48 a parallel worker. Depending on the join type, the inner side may also be49 a parallel plan.50 </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>51 In a <span class="emphasis"><em>nested loop join</em></span>, the inner side is always52 non-parallel. Although it is executed in full, this is efficient if53 the inner side is an index scan, because the outer tuples and thus54 the loops that look up values in the index are divided over the55 cooperating processes.56 </p></li><li class="listitem"><p>57 In a <span class="emphasis"><em>merge join</em></span>, the inner side is always58 a non-parallel plan and therefore executed in full. This may be59 inefficient, especially if a sort must be performed, because the work60 and resulting data are duplicated in every cooperating process.61 </p></li><li class="listitem"><p>62 In a <span class="emphasis"><em>hash join</em></span> (without the "parallel" prefix),63 the inner side is executed in full by every cooperating process64 to build identical copies of the hash table. This may be inefficient65 if the hash table is large or the plan is expensive. In a66 <span class="emphasis"><em>parallel hash join</em></span>, the inner side is a67 <span class="emphasis"><em>parallel hash</em></span> that divides the work of building68 a shared hash table over the cooperating processes.69 </p></li></ul></div></div><div class="sect2" id="PARALLEL-AGGREGATION"><div class="titlepage"><div><div><h3 class="title">15.3.3. Parallel Aggregation <a href="#PARALLEL-AGGREGATION" class="id_link">#</a></h3></div></div></div><p>70 <span class="productname">PostgreSQL</span> supports parallel aggregation by aggregating in71 two stages. First, each process participating in the parallel portion of72 the query performs an aggregation step, producing a partial result for73 each group of which that process is aware. This is reflected in the plan74 as a <code class="literal">Partial Aggregate</code> node. Second, the partial results are75 transferred to the leader via <code class="literal">Gather</code> or <code class="literal">Gather76 Merge</code>. Finally, the leader re-aggregates the results across all77 workers in order to produce the final result. This is reflected in the78 plan as a <code class="literal">Finalize Aggregate</code> node.79 </p><p>80 Because the <code class="literal">Finalize Aggregate</code> node runs on the leader81 process, queries that produce a relatively large number of groups in82 comparison to the number of input rows will appear less favorable to the83 query planner. For example, in the worst-case scenario the number of84 groups seen by the <code class="literal">Finalize Aggregate</code> node could be as many as85 the number of input rows that were seen by all worker processes in the86 <code class="literal">Partial Aggregate</code> stage. For such cases, there is clearly87 going to be no performance benefit to using parallel aggregation. The88 query planner takes this into account during the planning process and is89 unlikely to choose parallel aggregate in this scenario.90 </p><p>91 Parallel aggregation is not supported in all situations. Each aggregate92 must be <a class="link" href="parallel-safety.html" title="15.4. Parallel Safety">safe</a> for parallelism and must93 have a combine function. If the aggregate has a transition state of type94 <code class="literal">internal</code>, it must have serialization and deserialization95 functions. See <a class="xref" href="sql-createaggregate.html" title="CREATE AGGREGATE"><span class="refentrytitle">CREATE AGGREGATE</span></a> for more details.96 Parallel aggregation is not supported if any aggregate function call97 contains <code class="literal">DISTINCT</code> or <code class="literal">ORDER BY</code> clause and is also98 not supported for ordered set aggregates or when the query involves99 <code class="literal">GROUPING SETS</code>. It can only be used when all joins involved in100 the query are also part of the parallel portion of the plan.101 </p></div><div class="sect2" id="PARALLEL-APPEND"><div class="titlepage"><div><div><h3 class="title">15.3.4. Parallel Append <a href="#PARALLEL-APPEND" class="id_link">#</a></h3></div></div></div><p>102 Whenever <span class="productname">PostgreSQL</span> needs to combine rows103 from multiple sources into a single result set, it uses an104 <code class="literal">Append</code> or <code class="literal">MergeAppend</code> plan node.105 This commonly happens when implementing <code class="literal">UNION ALL</code> or106 when scanning a partitioned table. Such nodes can be used in parallel107 plans just as they can in any other plan. However, in a parallel plan,108 the planner may instead use a <code class="literal">Parallel Append</code> node.109 </p><p>110 When an <code class="literal">Append</code> node is used in a parallel plan, each111 process will execute the child plans in the order in which they appear,112 so that all participating processes cooperate to execute the first child113 plan until it is complete and then move to the second plan at around the114 same time. When a <code class="literal">Parallel Append</code> is used instead, the115 executor will instead spread out the participating processes as evenly as116 possible across its child plans, so that multiple child plans are executed117 simultaneously. This avoids contention, and also avoids paying the startup118 cost of a child plan in those processes that never execute it.119 </p><p>120 Also, unlike a regular <code class="literal">Append</code> node, which can only have121 partial children when used within a parallel plan, a <code class="literal">Parallel122 Append</code> node can have both partial and non-partial child plans.123 Non-partial children will be scanned by only a single process, since124 scanning them more than once would produce duplicate results. Plans that125 involve appending multiple results sets can therefore achieve126 coarse-grained parallelism even when efficient partial plans are not127 available. For example, consider a query against a partitioned table128 that can only be implemented efficiently by using an index that does129 not support parallel scans. The planner might choose a <code class="literal">Parallel130 Append</code> of regular <code class="literal">Index Scan</code> plans; each131 individual index scan would have to be executed to completion by a single132 process, but different scans could be performed at the same time by133 different processes.134 </p><p>135 <a class="xref" href="runtime-config-query.html#GUC-ENABLE-PARALLEL-APPEND">enable_parallel_append</a> can be used to disable136 this feature.137 </p></div><div class="sect2" id="PARALLEL-PLAN-TIPS"><div class="titlepage"><div><div><h3 class="title">15.3.5. Parallel Plan Tips <a href="#PARALLEL-PLAN-TIPS" class="id_link">#</a></h3></div></div></div><p>138 If a query that is expected to do so does not produce a parallel plan,139 you can try reducing <a class="xref" href="runtime-config-query.html#GUC-PARALLEL-SETUP-COST">parallel_setup_cost</a> or140 <a class="xref" href="runtime-config-query.html#GUC-PARALLEL-TUPLE-COST">parallel_tuple_cost</a>. Of course, this plan may turn141 out to be slower than the serial plan that the planner preferred, but142 this will not always be the case. If you don't get a parallel143 plan even with very small values of these settings (e.g., after setting144 them both to zero), there may be some reason why the query planner is145 unable to generate a parallel plan for your query. See146 <a class="xref" href="when-can-parallel-query-be-used.html" title="15.2. When Can Parallel Query Be Used?">Section 15.2</a> and147 <a class="xref" href="parallel-safety.html" title="15.4. Parallel Safety">Section 15.4</a> for information on why this may be148 the case.149 </p><p>150 When executing a parallel plan, you can use <code class="literal">EXPLAIN (ANALYZE,151 VERBOSE)</code> to display per-worker statistics for each plan node.152 This may be useful in determining whether the work is being evenly153 distributed between all plan nodes and more generally in understanding the154 performance characteristics of the plan.155 </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" title="15.2. When Can Parallel Query Be Used?">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="parallel-query.html" title="Chapter 15. Parallel Query">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="parallel-safety.html" title="15.4. Parallel Safety">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" title="PostgreSQL 16.3 Documentation">Home</a></td><td width="40%" align="right" valign="top"> 15.4. Parallel Safety</td></tr></table></div></body></html>