<?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>57.4. Foreign Data Wrapper Query Planning</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="fdw-helpers.html" title="57.3. Foreign Data Wrapper Helper Functions" /><link rel="next" href="fdw-row-locking.html" title="57.5. Row Locking in Foreign Data Wrappers" /></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">57.4. Foreign Data Wrapper Query Planning</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="fdw-helpers.html" title="57.3. Foreign Data Wrapper Helper Functions">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="fdwhandler.html" title="Chapter 57. Writing A Foreign Data Wrapper">Up</a></td><th width="60%" align="center">Chapter 57. Writing A Foreign Data Wrapper</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="fdw-row-locking.html" title="57.5. Row Locking in Foreign Data Wrappers">Next</a></td></tr></table><hr></hr></div><div class="sect1" id="FDW-PLANNING"><div class="titlepage"><div><div><h2 class="title" style="clear: both">57.4. Foreign Data Wrapper Query Planning</h2></div></div></div><p>
The FDW callback functions <code class="function">GetForeignRelSize</code>,
<code class="function">GetForeignPaths</code>, <code class="function">GetForeignPlan</code>,
<code class="function">PlanForeignModify</code>, <code class="function">GetForeignJoinPaths</code>,
<code class="function">GetForeignUpperPaths</code>, and <code class="function">PlanDirectModify</code>
must fit into the workings of the <span class="productname">PostgreSQL</span> planner.
Here are some notes about what they must do.
</p><p>
The information in <code class="literal">root</code> and <code class="literal">baserel</code> can be used
to reduce the amount of information that has to be fetched from the
foreign table (and therefore reduce the cost).
<code class="literal">baserel->baserestrictinfo</code> is particularly interesting, as
it contains restriction quals (<code class="literal">WHERE</code> clauses) that should be
used to filter the rows to be fetched. (The FDW itself is not required
to enforce these quals, as the core executor can check them instead.)
<code class="literal">baserel->reltarget->exprs</code> can be used to determine which
columns need to be fetched; but note that it only lists columns that
have to be emitted by the <code class="structname">ForeignScan</code> plan node, not
columns that are used in qual evaluation but not output by the query.
</p><p>
Various private fields are available for the FDW planning functions to
keep information in. Generally, whatever you store in FDW private fields
should be palloc'd, so that it will be reclaimed at the end of planning.
</p><p>
<code class="literal">baserel->fdw_private</code> is a <code class="type">void</code> pointer that is
available for FDW planning functions to store information relevant to
the particular foreign table. The core planner does not touch it except
to initialize it to NULL when the <code class="literal">RelOptInfo</code> node is created.
It is useful for passing information forward from
<code class="function">GetForeignRelSize</code> to <code class="function">GetForeignPaths</code> and/or
<code class="function">GetForeignPaths</code> to <code class="function">GetForeignPlan</code>, thereby
avoiding recalculation.
</p><p>
<code class="function">GetForeignPaths</code> can identify the meaning of different
access paths by storing private information in the
<code class="structfield">fdw_private</code> field of <code class="structname">ForeignPath</code> nodes.
<code class="structfield">fdw_private</code> is declared as a <code class="type">List</code> pointer, but
could actually contain anything since the core planner does not touch
it. However, best practice is to use a representation that's dumpable
by <code class="function">nodeToString</code>, for use with debugging support available
in the backend.
</p><p>
<code class="function">GetForeignPlan</code> can examine the <code class="structfield">fdw_private</code>
field of the selected <code class="structname">ForeignPath</code> node, and can generate
<code class="structfield">fdw_exprs</code> and <code class="structfield">fdw_private</code> lists to be
placed in the <code class="structname">ForeignScan</code> plan node, where they will be
available at execution time. Both of these lists must be
represented in a form that <code class="function">copyObject</code> knows how to copy.
The <code class="structfield">fdw_private</code> list has no other restrictions and is
not interpreted by the core backend in any way. The
<code class="structfield">fdw_exprs</code> list, if not NIL, is expected to contain
expression trees that are intended to be executed at run time. These
trees will undergo post-processing by the planner to make them fully
executable.
</p><p>
In <code class="function">GetForeignPlan</code>, generally the passed-in target list can
be copied into the plan node as-is. The passed <code class="literal">scan_clauses</code> list
contains the same clauses as <code class="literal">baserel->baserestrictinfo</code>,
but may be re-ordered for better execution efficiency. In simple cases
the FDW can just strip <code class="structname">RestrictInfo</code> nodes from the
<code class="literal">scan_clauses</code> list (using <code class="function">extract_actual_clauses</code>) and put
all the clauses into the plan node's qual list, which means that all the
clauses will be checked by the executor at run time. More complex FDWs
may be able to check some of the clauses internally, in which case those
clauses can be removed from the plan node's qual list so that the
executor doesn't waste time rechecking them.
</p><p>
As an example, the FDW might identify some restriction clauses of the
form <em class="replaceable"><code>foreign_variable</code></em> <code class="literal">=</code>
<em class="replaceable"><code>sub_expression</code></em>, which it determines can be executed on
the remote server given the locally-evaluated value of the
<em class="replaceable"><code>sub_expression</code></em>. The actual identification of such a
clause should happen during <code class="function">GetForeignPaths</code>, since it would
affect the cost estimate for the path. The path's
<code class="structfield">fdw_private</code> field would probably include a pointer to
the identified clause's <code class="structname">RestrictInfo</code> node. Then
<code class="function">GetForeignPlan</code> would remove that clause from <code class="literal">scan_clauses</code>,
but add the <em class="replaceable"><code>sub_expression</code></em> to <code class="structfield">fdw_exprs</code>
to ensure that it gets massaged into executable form. It would probably
also put control information into the plan node's
<code class="structfield">fdw_private</code> field to tell the execution functions what
to do at run time. The query transmitted to the remote server would
involve something like <code class="literal">WHERE <em class="replaceable"><code>foreign_variable</code></em> =
$1</code>, with the parameter value obtained at run time from
evaluation of the <code class="structfield">fdw_exprs</code> expression tree.
</p><p>
Any clauses removed from the plan node's qual list must instead be added
to <code class="literal">fdw_recheck_quals</code> or rechecked by
<code class="literal">RecheckForeignScan</code> in order to ensure correct behavior
at the <code class="literal">READ COMMITTED</code> isolation level. When a concurrent
update occurs for some other table involved in the query, the executor
may need to verify that all of the original quals are still satisfied for
the tuple, possibly against a different set of parameter values. Using
<code class="literal">fdw_recheck_quals</code> is typically easier than implementing checks
inside <code class="literal">RecheckForeignScan</code>, but this method will be
insufficient when outer joins have been pushed down, since the join tuples
in that case might have some fields go to NULL without rejecting the
tuple entirely.
</p><p>
Another <code class="structname">ForeignScan</code> field that can be filled by FDWs
is <code class="structfield">fdw_scan_tlist</code>, which describes the tuples returned by
the FDW for this plan node. For simple foreign table scans this can be
set to <code class="literal">NIL</code>, implying that the returned tuples have the
row type declared for the foreign table. A non-<code class="symbol">NIL</code> value must be a
target list (list of <code class="structname">TargetEntry</code>s) containing Vars and/or
expressions representing the returned columns. This might be used, for
example, to show that the FDW has omitted some columns that it noticed
won't be needed for the query. Also, if the FDW can compute expressions
used by the query more cheaply than can be done locally, it could add
those expressions to <code class="structfield">fdw_scan_tlist</code>. Note that join
plans (created from paths made by <code class="function">GetForeignJoinPaths</code>) must
always supply <code class="structfield">fdw_scan_tlist</code> to describe the set of
columns they will return.
</p><p>
The FDW should always construct at least one path that depends only on
the table's restriction clauses. In join queries, it might also choose
to construct path(s) that depend on join clauses, for example
<em class="replaceable"><code>foreign_variable</code></em> <code class="literal">=</code>
<em class="replaceable"><code>local_variable</code></em>. Such clauses will not be found in
<code class="literal">baserel->baserestrictinfo</code> but must be sought in the
relation's join lists. A path using such a clause is called a
<span class="quote">“<span class="quote">parameterized path</span>”</span>. It must identify the other relations
used in the selected join clause(s) with a suitable value of
<code class="literal">param_info</code>; use <code class="function">get_baserel_parampathinfo</code>
to compute that value. In <code class="function">GetForeignPlan</code>, the
<em class="replaceable"><code>local_variable</code></em> portion of the join clause would be added
to <code class="structfield">fdw_exprs</code>, and then at run time the case works the
same as for an ordinary restriction clause.
</p><p>
If an FDW supports remote joins, <code class="function">GetForeignJoinPaths</code> should
produce <code class="structname">ForeignPath</code>s for potential remote joins in much
the same way as <code class="function">GetForeignPaths</code> works for base tables.
Information about the intended join can be passed forward
to <code class="function">GetForeignPlan</code> in the same ways described above.
However, <code class="structfield">baserestrictinfo</code> is not relevant for join
relations; instead, the relevant join clauses for a particular join are
passed to <code class="function">GetForeignJoinPaths</code> as a separate parameter
(<code class="literal">extra->restrictlist</code>).
</p><p>
An FDW might additionally support direct execution of some plan actions
that are above the level of scans and joins, such as grouping or
aggregation. To offer such options, the FDW should generate paths and
insert them into the appropriate <em class="firstterm">upper relation</em>. For
example, a path representing remote aggregation should be inserted into
the <code class="literal">UPPERREL_GROUP_AGG</code> relation, using <code class="function">add_path</code>.
This path will be compared on a cost basis with local aggregation
performed by reading a simple scan path for the foreign relation (note
that such a path must also be supplied, else there will be an error at
plan time). If the remote-aggregation path wins, which it usually would,
it will be converted into a plan in the usual way, by
calling <code class="function">GetForeignPlan</code>. The recommended place to generate
such paths is in the <code class="function">GetForeignUpperPaths</code>
callback function, which is called for each upper relation (i.e., each
post-scan/join processing step), if all the base relations of the query
come from the same FDW.
</p><p>
<code class="function">PlanForeignModify</code> and the other callbacks described in
<a class="xref" href="fdw-callbacks.html#FDW-CALLBACKS-UPDATE" title="57.2.4. FDW Routines For Updating Foreign Tables">Section 57.2.4</a> are designed around the assumption
that the foreign relation will be scanned in the usual way and then
individual row updates will be driven by a local <code class="literal">ModifyTable</code>
plan node. This approach is necessary for the general case where an
update requires reading local tables as well as foreign tables.
However, if the operation could be executed entirely by the foreign
server, the FDW could generate a path representing that and insert it
into the <code class="literal">UPPERREL_FINAL</code> upper relation, where it would
compete against the <code class="literal">ModifyTable</code> approach. This approach
could also be used to implement remote <code class="literal">SELECT FOR UPDATE</code>,
rather than using the row locking callbacks described in
<a class="xref" href="fdw-callbacks.html#FDW-CALLBACKS-ROW-LOCKING" title="57.2.5. FDW Routines For Row Locking">Section 57.2.5</a>. Keep in mind that a path
inserted into <code class="literal">UPPERREL_FINAL</code> is responsible for
implementing <span class="emphasis"><em>all</em></span> behavior of the query.
</p><p>
When planning an <code class="command">UPDATE</code> or <code class="command">DELETE</code>,
<code class="function">PlanForeignModify</code> and <code class="function">PlanDirectModify</code>
can look up the <code class="structname">RelOptInfo</code>
struct for the foreign table and make use of the
<code class="literal">baserel->fdw_private</code> data previously created by the
scan-planning functions. However, in <code class="command">INSERT</code> the target
table is not scanned so there is no <code class="structname">RelOptInfo</code> for it.
The <code class="structname">List</code> returned by <code class="function">PlanForeignModify</code> has
the same restrictions as the <code class="structfield">fdw_private</code> list of a
<code class="structname">ForeignScan</code> plan node, that is it must contain only
structures that <code class="function">copyObject</code> knows how to copy.
</p><p>
<code class="command">INSERT</code> with an <code class="literal">ON CONFLICT</code> clause does not
support specifying the conflict target, as unique constraints or
exclusion constraints on remote tables are not locally known. This
in turn implies that <code class="literal">ON CONFLICT DO UPDATE</code> is not supported,
since the specification is mandatory there.
</p></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="fdw-helpers.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="fdwhandler.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="fdw-row-locking.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">57.3. Foreign Data Wrapper Helper Functions </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> 57.5. Row Locking in Foreign Data Wrappers</td></tr></table></div></body></html>
