Add a GUC parameter seq_page_cost, and use that everywhere we formerly

assumed that a sequential page fetch has cost 1.0.  This patch doesn't
in itself change the system's behavior at all, but it opens the door to
people adopting other units of measurement for EXPLAIN costs.  Also, if
we ever decide it's worth inventing per-tablespace access cost settings,
this change provides a workable intellectual framework for that.

doc/src/sgml/config.sgml

-<!-- $PostgreSQL: pgsql/doc/src/sgml/config.sgml,v 1.59 2006/05/21 20:10:42 tgl Exp $ -->
+<!-- $PostgreSQL: pgsql/doc/src/sgml/config.sgml,v 1.60 2006/06/05 02:49:58 tgl Exp $ -->
 
 <chapter Id="runtime-config">
   <title>Server Configuration</title>
@@ -1739,40 +1739,39 @@ archive_command = 'copy "%p" /mnt/server/archivedir/"%f"'  # Windows
       Planner Cost Constants
      </title>
 
+    <para>
+     The <firstterm>cost</> variables described in this section are measured
+     on an arbitrary scale.  Only their relative values matter, hence
+     scaling them all up or down by the same factor will result in no change
+     in the planner's choices.  Traditionally, these variables have been
+     referenced to sequential page fetches as the unit of cost; that is,
+     <varname>seq_page_cost</> is conventionally set to <literal>1.0</>
+     and the other cost variables are set with reference to that.  But
+     you can use a different scale if you prefer, such as actual execution
+     times in milliseconds on a particular machine.
+    </para>
+
    <note>
     <para>
-     Unfortunately, there is no well-defined method for determining
-     ideal values for the family of <quote>cost</quote> variables that
-     appear below. You are encouraged to experiment and share
-     your findings.
+     Unfortunately, there is no well-defined method for determining ideal
+     values for the cost variables.  They are best treated as averages over
+     the entire mix of queries that a particular installation will get.  This
+     means that changing them on the basis of just a few experiments is very
+     risky.
     </para>
    </note>
 
      <variablelist>
-     
-     <varlistentry id="guc-effective-cache-size" xreflabel="effective_cache_size">
-      <term><varname>effective_cache_size</varname> (<type>floating point</type>)</term>
+
+     <varlistentry id="guc-seq-page-cost" xreflabel="seq_page_cost">
+      <term><varname>seq_page_cost</varname> (<type>floating point</type>)</term>
       <indexterm>
-       <primary><varname>effective_cache_size</> configuration parameter</primary>
+       <primary><varname>seq_page_cost</> configuration parameter</primary>
       </indexterm>
       <listitem>
        <para>
-        Sets the planner's assumption about the effective size of the
-        disk cache that is available to a single index scan.  This is
-        factored into estimates of the cost of using an index; a
-        higher value makes it more likely index scans will be used, a
-        lower value makes it more likely sequential scans will be
-        used. When setting this parameter you should consider both
-        <productname>PostgreSQL</productname>'s shared buffers and the
-        portion of the kernel's disk cache that will be used for
-        <productname>PostgreSQL</productname> data files.  Also, take
-        into account the expected number of concurrent queries using
-        different indexes, since they will have to share the available
-        space.  This parameter has no effect on the size of shared
-        memory allocated by <productname>PostgreSQL</productname>, nor
-        does it reserve kernel disk cache; it is used only for
-        estimation purposes.  The value is measured in disk pages,
-        which are normally 8192 bytes each. The default is 1000.
+        Sets the planner's estimate of the cost of a disk page fetch
+        that is part of a series of sequential fetches.  The default is 1.0.
        </para>
       </listitem>
      </varlistentry>
@@ -1785,12 +1784,27 @@ archive_command = 'copy "%p" /mnt/server/archivedir/"%f"'  # Windows
       <listitem>
        <para>
         Sets the planner's estimate of the cost of a
-        nonsequentially fetched disk page. This is measured as a
-        multiple of the cost of a sequential page fetch. A higher
-        value makes it more likely a sequential scan will be used, a
-        lower value makes it more likely an index scan will be
-        used. The default is four.
+        non-sequentially-fetched disk page.  The default is 4.0.
+        Reducing this value relative to <varname>seq_page_cost</>
+        will cause the system to prefer index scans; raising it will
+        make index scans look relatively more expensive.  You can raise
+        or lower both values together to change the importance of disk I/O
+        costs relative to CPU costs, which are described by the following
+        parameters.
        </para>
+
+       <tip>
+        <para>
+         Although the system will let you set <varname>random_page_cost</> to
+         less than <varname>seq_page_cost</>, it is not physically sensible
+         to do so.  However, setting them equal makes sense if the database
+         is entirely cached in RAM, since in that case there is no penalty
+         for touching pages out of sequence.  Also, in a heavily-cached
+         database you should lower both values relative to the CPU parameters,
+         since the cost of fetching a page already in RAM is much smaller
+         than it would normally be.
+        </para>
+       </tip>
       </listitem>
      </varlistentry>
 
@@ -1802,8 +1816,8 @@ archive_command = 'copy "%p" /mnt/server/archivedir/"%f"'  # Windows
       <listitem>
        <para>
         Sets the planner's estimate of the cost of processing
-        each row during a query. This is measured as a fraction of
-        the cost of a sequential page fetch. The default is 0.01.
+        each row during a query.
+        The default is 0.01.
        </para>
       </listitem>
      </varlistentry>
@@ -1816,9 +1830,8 @@ archive_command = 'copy "%p" /mnt/server/archivedir/"%f"'  # Windows
       <listitem>
        <para>
         Sets the planner's estimate of the cost of processing
-        each index row during an index scan. This is measured as a
-        fraction of the cost of a sequential page fetch. The default
-        is 0.001.
+        each index entry during an index scan.
+        The default is 0.001.
        </para>
       </listitem>
      </varlistentry>
@@ -1831,8 +1844,35 @@ archive_command = 'copy "%p" /mnt/server/archivedir/"%f"'  # Windows
       <listitem>
        <para>
         Sets the planner's estimate of the cost of processing each
-        operator in a <literal>WHERE</> clause. This is measured as a fraction of
-        the cost of a sequential page fetch. The default is 0.0025.
+        operator or function executed during a query.
+        The default is 0.0025.
+       </para>
+      </listitem>
+     </varlistentry>
+     
+     <varlistentry id="guc-effective-cache-size" xreflabel="effective_cache_size">
+      <term><varname>effective_cache_size</varname> (<type>floating point</type>)</term>
+      <indexterm>
+       <primary><varname>effective_cache_size</> configuration parameter</primary>
+      </indexterm>
+      <listitem>
+       <para>
+        Sets the planner's assumption about the effective size of the
+        disk cache that is available to a single index scan.  This is
+        factored into estimates of the cost of using an index; a
+        higher value makes it more likely index scans will be used, a
+        lower value makes it more likely sequential scans will be
+        used. When setting this parameter you should consider both
+        <productname>PostgreSQL</productname>'s shared buffers and the
+        portion of the kernel's disk cache that will be used for
+        <productname>PostgreSQL</productname> data files.  Also, take
+        into account the expected number of concurrent queries using
+        different indexes, since they will have to share the available
+        space.  This parameter has no effect on the size of shared
+        memory allocated by <productname>PostgreSQL</productname>, nor
+        does it reserve kernel disk cache; it is used only for
+        estimation purposes.  The value is measured in disk pages,
+        which are normally 8192 bytes each. The default is 1000.
        </para>
       </listitem>
      </varlistentry>

doc/src/sgml/indexam.sgml

-<!-- $PostgreSQL: pgsql/doc/src/sgml/indexam.sgml,v 2.12 2006/05/24 11:01:39 teodor Exp $ -->
+<!-- $PostgreSQL: pgsql/doc/src/sgml/indexam.sgml,v 2.13 2006/06/05 02:49:58 tgl Exp $ -->
 
 <chapter id="indexam">
  <title>Index Access Method Interface Definition</title>
@@ -771,14 +771,14 @@ amcostestimate (PlannerInfo *root,
   </para>
 
   <para>
-   The index access costs should be computed in the units used by
+   The index access costs should be computed using the parameters used by
    <filename>src/backend/optimizer/path/costsize.c</filename>: a sequential
-   disk block fetch has cost 1.0, a nonsequential fetch has cost
-   <varname>random_page_cost</>, and the cost of processing one index row
-   should usually be taken as <varname>cpu_index_tuple_cost</>.  In addition,
-   an appropriate multiple of <varname>cpu_operator_cost</> should be charged
-   for any comparison operators invoked during index processing (especially
-   evaluation of the indexQuals themselves).
+   disk block fetch has cost <varname>seq_page_cost</>, a nonsequential fetch
+   has cost <varname>random_page_cost</>, and the cost of processing one index
+   row should usually be taken as <varname>cpu_index_tuple_cost</>.  In
+   addition, an appropriate multiple of <varname>cpu_operator_cost</> should
+   be charged for any comparison operators invoked during index processing
+   (especially evaluation of the indexQuals themselves).
   </para>
 
   <para>
@@ -788,10 +788,10 @@ amcostestimate (PlannerInfo *root,
   </para>
 
   <para>
-   The <quote>start-up cost</quote> is the part of the total scan cost that must be expended
-   before we can begin to fetch the first row.  For most indexes this can
-   be taken as zero, but an index type with a high start-up cost might want
-   to set it nonzero.
+   The <quote>start-up cost</quote> is the part of the total scan cost that
+   must be expended before we can begin to fetch the first row.  For most
+   indexes this can be taken as zero, but an index type with a high start-up
+   cost might want to set it nonzero.
   </para>
 
   <para>
@@ -850,13 +850,13 @@ amcostestimate (PlannerInfo *root,
 <programlisting>
     /*
      * Our generic assumption is that the index pages will be read
-     * sequentially, so they have cost 1.0 each, not random_page_cost.
+     * sequentially, so they cost seq_page_cost each, not random_page_cost.
      * Also, we charge for evaluation of the indexquals at each index row.
      * All the costs are assumed to be paid incrementally during the scan.
      */
     cost_qual_eval(&amp;index_qual_cost, indexQuals);
     *indexStartupCost = index_qual_cost.startup;
-    *indexTotalCost = numIndexPages +
+    *indexTotalCost = seq_page_cost * numIndexPages +
         (cpu_index_tuple_cost + index_qual_cost.per_tuple) * numIndexTuples;
 </programlisting>
     </para>

doc/src/sgml/perform.sgml

-<!-- $PostgreSQL: pgsql/doc/src/sgml/perform.sgml,v 1.56 2006/03/10 19:10:48 momjian Exp $ -->
+<!-- $PostgreSQL: pgsql/doc/src/sgml/perform.sgml,v 1.57 2006/06/05 02:49:58 tgl Exp $ -->
 
  <chapter id="performance-tips">
   <title>Performance Tips</title>
@@ -60,7 +60,7 @@
     <footnote>
      <para>
       Examples in this section are drawn from the regression test database
-      after doing a <command>VACUUM ANALYZE</>, using 8.1 development sources.
+      after doing a <command>VACUUM ANALYZE</>, using 8.2 development sources.
       You should be able to get similar results if you try the examples yourself,
       but your estimated costs and row counts will probably vary slightly
       because <command>ANALYZE</>'s statistics are random samples rather
@@ -114,12 +114,13 @@ EXPLAIN SELECT * FROM tenk1;
    </para>
 
    <para>
-    The costs are measured in units of disk page fetches; that is, 1.0
-    equals one sequential disk page read, by definition.  (CPU effort
-    estimates are made too; they are converted into disk-page units using some
-    fairly arbitrary fudge factors. If you want to experiment with these
-    factors, see the list of run-time configuration parameters in
-    <xref linkend="runtime-config-query-constants">.)
+    The costs are measured in arbitrary units determined by the planner's
+    cost parameters (see <xref linkend="runtime-config-query-constants">).
+    Traditional practice is to measure the costs in units of disk page
+    fetches; that is, <xref linkend="guc-seq-page-cost"> is conventionally
+    set to <literal>1.0</> and the other cost parameters are set relative
+    to that.  The examples in this section are run with the default cost
+    parameters.
    </para>
 
    <para>
@@ -164,9 +165,9 @@ SELECT relpages, reltuples FROM pg_class WHERE relname = 'tenk1';
 
     you will find out that <classname>tenk1</classname> has 358 disk
     pages and 10000 rows.  So the cost is estimated at 358 page
-    reads, defined as costing 1.0 apiece, plus 10000 * <xref
-    linkend="guc-cpu-tuple-cost"> which is
-    typically 0.01 (try <command>SHOW cpu_tuple_cost</command>).
+    reads, costing <xref linkend="guc-seq-page-cost"> apiece (1.0 by
+    default), plus 10000 * <xref linkend="guc-cpu-tuple-cost"> which is
+    0.01 by default.
    </para>
 
    <para>
@@ -400,8 +401,9 @@ EXPLAIN ANALYZE SELECT * FROM tenk1 t1, tenk2 t2 WHERE t1.unique1 &lt; 100 AND t
 
     Note that the <quote>actual time</quote> values are in milliseconds of
     real time, whereas the <quote>cost</quote> estimates are expressed in
-    arbitrary units of disk fetches; so they are unlikely to match up.
-    The thing to pay attention to is the ratios.
+    arbitrary units; so they are unlikely to match up.
+    The thing to pay attention to is whether the ratios of actual time and
+    estimated costs are consistent.
    </para>
 
    <para>
@@ -427,7 +429,7 @@ EXPLAIN ANALYZE SELECT * FROM tenk1 t1, tenk2 t2 WHERE t1.unique1 &lt; 100 AND t
     may be considerably larger, because it includes the time spent processing
     the result rows.  In these commands, the time for the top plan node
     essentially is the time spent computing the new rows and/or locating the
-    old ones, but it doesn't include the time spent making the changes.
+    old ones, but it doesn't include the time spent applying the changes.
     Time spent firing triggers, if any, is also outside the top plan node,
     and is shown separately for each trigger.
    </para>

src/backend/optimizer/path/costsize.c

@@ -3,14 +3,19 @@
  * costsize.c
  *	  Routines to compute (and set) relation sizes and path costs
  *
- * Path costs are measured in units of disk accesses: one sequential page
- * fetch has cost 1.  All else is scaled relative to a page fetch, using
- * the scaling parameters
+ * Path costs are measured in arbitrary units established by these basic
+ * parameters:
  *
+ *	seq_page_cost		Cost of a sequential page fetch
  *	random_page_cost	Cost of a non-sequential page fetch
  *	cpu_tuple_cost		Cost of typical CPU time to process a tuple
  *	cpu_index_tuple_cost  Cost of typical CPU time to process an index tuple
- *	cpu_operator_cost	Cost of CPU time to process a typical WHERE operator
+ *	cpu_operator_cost	Cost of CPU time to execute an operator or function
+ *
+ * We expect that the kernel will typically do some amount of read-ahead
+ * optimization; this in conjunction with seek costs means that seq_page_cost
+ * is normally considerably less than random_page_cost.  (However, if the
+ * database is fully cached in RAM, it is reasonable to set them equal.)
  *
  * We also use a rough estimate "effective_cache_size" of the number of
  * disk pages in Postgres + OS-level disk cache.  (We can't simply use
@@ -49,7 +54,7 @@
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/optimizer/path/costsize.c,v 1.155 2006/03/05 15:58:28 momjian Exp $
+ *	  $PostgreSQL: pgsql/src/backend/optimizer/path/costsize.c,v 1.156 2006/06/05 02:49:58 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -85,12 +90,14 @@
 	 (path)->parent->rows)
 
 
-double		effective_cache_size = DEFAULT_EFFECTIVE_CACHE_SIZE;
+double		seq_page_cost = DEFAULT_SEQ_PAGE_COST;
 double		random_page_cost = DEFAULT_RANDOM_PAGE_COST;
 double		cpu_tuple_cost = DEFAULT_CPU_TUPLE_COST;
 double		cpu_index_tuple_cost = DEFAULT_CPU_INDEX_TUPLE_COST;
 double		cpu_operator_cost = DEFAULT_CPU_OPERATOR_COST;
 
+double		effective_cache_size = DEFAULT_EFFECTIVE_CACHE_SIZE;
+
 Cost		disable_cost = 100000000.0;
 
 bool		enable_seqscan = true;
@@ -156,14 +163,8 @@ cost_seqscan(Path *path, PlannerInfo *root,
 
 	/*
 	 * disk costs
-	 *
-	 * The cost of reading a page sequentially is 1.0, by definition. Note
-	 * that the Unix kernel will typically do some amount of read-ahead
-	 * optimization, so that this cost is less than the true cost of reading a
-	 * page from disk.	We ignore that issue here, but must take it into
-	 * account when estimating the cost of non-sequential accesses!
 	 */
-	run_cost += baserel->pages; /* sequential fetches with cost 1.0 */
+	run_cost += seq_page_cost * baserel->pages;
 
 	/* CPU costs */
 	startup_cost += baserel->baserestrictcost.startup;
@@ -194,20 +195,21 @@ cost_seqscan(Path *path, PlannerInfo *root,
  * with the entirely ad-hoc equations (writing relsize for
  * relpages/effective_cache_size):
  *	if relsize >= 1:
- *		random_page_cost - (random_page_cost-1)/2 * (1/relsize)
+ *		random_page_cost - (random_page_cost-seq_page_cost)/2 * (1/relsize)
  *	if relsize < 1:
- *		1 + ((random_page_cost-1)/2) * relsize ** 2
- * These give the right asymptotic behavior (=> 1.0 as relpages becomes
- * small, => random_page_cost as it becomes large) and meet in the middle
- * with the estimate that the cache is about 50% effective for a relation
- * of the same size as effective_cache_size.  (XXX this is probably all
- * wrong, but I haven't been able to find any theory about how effective
+ *		seq_page_cost + ((random_page_cost-seq_page_cost)/2) * relsize ** 2
+ * These give the right asymptotic behavior (=> seq_page_cost as relpages
+ * becomes small, => random_page_cost as it becomes large) and meet in the
+ * middle with the estimate that the cache is about 50% effective for a
+ * relation of the same size as effective_cache_size.  (XXX this is probably
+ * all wrong, but I haven't been able to find any theory about how effective
  * a disk cache should be presumed to be.)
  */
 static Cost
 cost_nonsequential_access(double relpages)
 {
 	double		relsize;
+	double		random_delta;
 
 	/* don't crash on bad input data */
 	if (relpages <= 0.0 || effective_cache_size <= 0.0)
@@ -215,19 +217,17 @@ cost_nonsequential_access(double relpages)
 
 	relsize = relpages / effective_cache_size;
 
+	random_delta = (random_page_cost - seq_page_cost) * 0.5;
 	if (relsize >= 1.0)
-		return random_page_cost - (random_page_cost - 1.0) * 0.5 / relsize;
+		return random_page_cost - random_delta / relsize;
 	else
-		return 1.0 + (random_page_cost - 1.0) * 0.5 * relsize * relsize;
+		return seq_page_cost + random_delta * relsize * relsize;
 }
 
 /*
  * cost_index
  *	  Determines and returns the cost of scanning a relation using an index.
  *
- *	  NOTE: an indexscan plan node can actually represent several passes,
- *	  but here we consider the cost of just one pass.
- *
  * 'index' is the index to be used
  * 'indexQuals' is the list of applicable qual clauses (implicit AND semantics)
  * 'is_injoin' is T if we are considering using the index scan as the inside
@@ -327,9 +327,9 @@ cost_index(IndexPath *path, PlannerInfo *root,
 	 * be just sT.	What's more, these will be sequential fetches, not the
 	 * random fetches that occur in the uncorrelated case.	So, depending on
 	 * the extent of correlation, we should estimate the actual I/O cost
-	 * somewhere between s * T * 1.0 and PF * random_cost.	We currently
-	 * interpolate linearly between these two endpoints based on the
-	 * correlation squared (XXX is that appropriate?).
+	 * somewhere between s * T * seq_page_cost and PF * random_page_cost.
+	 * We currently interpolate linearly between these two endpoints based on
+	 * the correlation squared (XXX is that appropriate?).
 	 *
 	 * In any case the number of tuples fetched is Ns.
 	 *----------
@@ -346,8 +346,10 @@ cost_index(IndexPath *path, PlannerInfo *root,
 	{
 		pages_fetched =
 			(2.0 * T * tuples_fetched) / (2.0 * T + tuples_fetched);
-		if (pages_fetched > T)
+		if (pages_fetched >= T)
 			pages_fetched = T;
+		else
+			pages_fetched = ceil(pages_fetched);
 	}
 	else
 	{
@@ -364,6 +366,7 @@ cost_index(IndexPath *path, PlannerInfo *root,
 			pages_fetched =
 				b + (tuples_fetched - lim) * (T - b) / T;
 		}
+		pages_fetched = ceil(pages_fetched);
 	}
 
 	/*
@@ -373,7 +376,7 @@ cost_index(IndexPath *path, PlannerInfo *root,
 	 * rather than using cost_nonsequential_access, since we've already
 	 * accounted for caching effects by using the Mackert model.
 	 */
-	min_IO_cost = ceil(indexSelectivity * T);
+	min_IO_cost = ceil(indexSelectivity * T) * seq_page_cost;
 	max_IO_cost = pages_fetched * random_page_cost;
 
 	/*
@@ -461,19 +464,21 @@ cost_bitmap_heap_scan(Path *path, PlannerInfo *root, RelOptInfo *baserel,
 
 	T = (baserel->pages > 1) ? (double) baserel->pages : 1.0;
 	pages_fetched = (2.0 * T * tuples_fetched) / (2.0 * T + tuples_fetched);
-	if (pages_fetched > T)
+	if (pages_fetched >= T)
 		pages_fetched = T;
+	else
+		pages_fetched = ceil(pages_fetched);
 
 	/*
 	 * For small numbers of pages we should charge random_page_cost apiece,
 	 * while if nearly all the table's pages are being read, it's more
-	 * appropriate to charge 1.0 apiece.  The effect is nonlinear, too. For
-	 * lack of a better idea, interpolate like this to determine the cost per
-	 * page.
+	 * appropriate to charge seq_page_cost apiece.  The effect is nonlinear,
+	 * too. For lack of a better idea, interpolate like this to determine the
+	 * cost per page.
 	 */
 	if (pages_fetched >= 2.0)
 		cost_per_page = random_page_cost -
-			(random_page_cost - 1.0) * sqrt(pages_fetched / T);
+			(random_page_cost - seq_page_cost) * sqrt(pages_fetched / T);
 	else
 		cost_per_page = random_page_cost;
 
@@ -833,9 +838,9 @@ cost_sort(Path *path, PlannerInfo *root,
 		else
 			log_runs = 1.0;
 		npageaccesses = 2.0 * npages * log_runs;
-		/* Assume half are sequential (cost 1), half are not */
+		/* Assume half are sequential, half are not */
 		startup_cost += npageaccesses *
-			(1.0 + cost_nonsequential_access(npages)) * 0.5;
+			(seq_page_cost + cost_nonsequential_access(npages)) * 0.5;
 	}
 
 	/*
@@ -871,8 +876,8 @@ cost_material(Path *path,
 		double		npages = ceil(nbytes / BLCKSZ);
 
 		/* We'll write during startup and read during retrieval */
-		startup_cost += npages;
-		run_cost += npages;
+		startup_cost += seq_page_cost * npages;
+		run_cost += seq_page_cost * npages;
 	}
 
 	/*

src/backend/utils/adt/selfuncs.c

@@ -15,7 +15,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/utils/adt/selfuncs.c,v 1.205 2006/05/02 11:28:55 teodor Exp $
+ *	  $PostgreSQL: pgsql/src/backend/utils/adt/selfuncs.c,v 1.206 2006/06/05 02:49:58 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -4555,9 +4555,9 @@ genericcostestimate(PlannerInfo *root,
 	 * Compute the index access cost.
 	 *
 	 * Disk cost: our generic assumption is that the index pages will be read
-	 * sequentially, so they have cost 1.0 each, not random_page_cost.
+	 * sequentially, so they cost seq_page_cost each, not random_page_cost.
 	 */
-	*indexTotalCost = numIndexPages;
+	*indexTotalCost = seq_page_cost * numIndexPages;
 
 	/*
 	 * CPU cost: any complex expressions in the indexquals will need to be

src/backend/utils/misc/guc.c

@@ -10,7 +10,7 @@
  * Written by Peter Eisentraut <peter_e@gmx.net>.
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/utils/misc/guc.c,v 1.320 2006/05/21 20:10:42 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/utils/misc/guc.c,v 1.321 2006/06/05 02:49:58 tgl Exp $
  *
  *--------------------------------------------------------------------
  */
@@ -1595,56 +1595,62 @@ static struct config_int ConfigureNamesInt[] =
 static struct config_real ConfigureNamesReal[] =
 {
 	{
-		{"effective_cache_size", PGC_USERSET, QUERY_TUNING_COST,
-			gettext_noop("Sets the planner's assumption about size of the disk cache."),
-			gettext_noop("That is, the portion of the kernel's disk cache that "
-						 "will be used for PostgreSQL data files. This is measured in disk "
-						 "pages, which are normally 8 kB each.")
+		{"seq_page_cost", PGC_USERSET, QUERY_TUNING_COST,
+			gettext_noop("Sets the planner's estimate of the cost of a "
+						 "sequentially fetched disk page."),
+			NULL
 		},
-		&effective_cache_size,
-		DEFAULT_EFFECTIVE_CACHE_SIZE, 1, DBL_MAX, NULL, NULL
+		&seq_page_cost,
+		DEFAULT_SEQ_PAGE_COST, 0, DBL_MAX, NULL, NULL
 	},
 	{
 		{"random_page_cost", PGC_USERSET, QUERY_TUNING_COST,
-			gettext_noop("Sets the planner's estimate of the cost of a nonsequentially "
-						 "fetched disk page."),
-			gettext_noop("This is measured as a multiple of the cost of a "
-			  "sequential page fetch. A higher value makes it more likely a "
-						 "sequential scan will be used, a lower value makes it more likely an "
-						 "index scan will be used.")
+			gettext_noop("Sets the planner's estimate of the cost of a "
+						 "nonsequentially fetched disk page."),
+			NULL
 		},
 		&random_page_cost,
 		DEFAULT_RANDOM_PAGE_COST, 0, DBL_MAX, NULL, NULL
 	},
 	{
 		{"cpu_tuple_cost", PGC_USERSET, QUERY_TUNING_COST,
-			gettext_noop("Sets the planner's estimate of the cost of processing each tuple (row)."),
-			gettext_noop("This is measured as a fraction of the cost of a "
-						 "sequential page fetch.")
+			gettext_noop("Sets the planner's estimate of the cost of "
+						 "processing each tuple (row)."),
+			NULL
 		},
 		&cpu_tuple_cost,
 		DEFAULT_CPU_TUPLE_COST, 0, DBL_MAX, NULL, NULL
 	},
 	{
 		{"cpu_index_tuple_cost", PGC_USERSET, QUERY_TUNING_COST,
-			gettext_noop("Sets the planner's estimate of processing cost for each "
-						 "index tuple (row) during index scan."),
-			gettext_noop("This is measured as a fraction of the cost of a "
-						 "sequential page fetch.")
+			gettext_noop("Sets the planner's estimate of the cost of "
+						 "processing each index entry during an index scan."),
+			NULL
 		},
 		&cpu_index_tuple_cost,
 		DEFAULT_CPU_INDEX_TUPLE_COST, 0, DBL_MAX, NULL, NULL
 	},
 	{
 		{"cpu_operator_cost", PGC_USERSET, QUERY_TUNING_COST,
-			gettext_noop("Sets the planner's estimate of processing cost of each operator in WHERE."),
-			gettext_noop("This is measured as a fraction of the cost of a sequential "
-						 "page fetch.")
+			gettext_noop("Sets the planner's estimate of the cost of "
+						 "processing each operator or function call."),
+			NULL
 		},
 		&cpu_operator_cost,
 		DEFAULT_CPU_OPERATOR_COST, 0, DBL_MAX, NULL, NULL
 	},
 
+	{
+		{"effective_cache_size", PGC_USERSET, QUERY_TUNING_COST,
+			gettext_noop("Sets the planner's assumption about size of the disk cache."),
+			gettext_noop("That is, the portion of the kernel's disk cache that "
+						 "will be used for PostgreSQL data files. This is measured in disk "
+						 "pages, which are normally 8 kB each.")
+		},
+		&effective_cache_size,
+		DEFAULT_EFFECTIVE_CACHE_SIZE, 1, DBL_MAX, NULL, NULL
+	},
+
 	{
 		{"geqo_selection_bias", PGC_USERSET, QUERY_TUNING_GEQO,
 			gettext_noop("GEQO: selective pressure within the population."),

src/backend/utils/misc/postgresql.conf.sample

@@ -175,12 +175,12 @@
 
 # - Planner Cost Constants -
 
+#seq_page_cost = 1.0			# measured on an arbitrary scale
+#random_page_cost = 4.0			# same scale as above
+#cpu_tuple_cost = 0.01			# same scale as above
+#cpu_index_tuple_cost = 0.001		# same scale as above
+#cpu_operator_cost = 0.0025		# same scale as above
 #effective_cache_size = 1000		# typically 8KB each
-#random_page_cost = 4			# units are one sequential page fetch 
-					# cost
-#cpu_tuple_cost = 0.01			# (same)
-#cpu_index_tuple_cost = 0.001		# (same)
-#cpu_operator_cost = 0.0025		# (same)
 
 # - Genetic Query Optimizer -
 

src/include/optimizer/cost.h

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2006, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/optimizer/cost.h,v 1.73 2006/03/05 15:58:57 momjian Exp $
+ * $PostgreSQL: pgsql/src/include/optimizer/cost.h,v 1.74 2006/06/05 02:49:58 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -21,12 +21,14 @@
 /* defaults for costsize.c's Cost parameters */
 /* NB: cost-estimation code should use the variables, not these constants! */
 /* If you change these, update backend/utils/misc/postgresql.sample.conf */
-#define DEFAULT_EFFECTIVE_CACHE_SIZE  1000.0	/* measured in pages */
+#define DEFAULT_SEQ_PAGE_COST  1.0
 #define DEFAULT_RANDOM_PAGE_COST  4.0
 #define DEFAULT_CPU_TUPLE_COST	0.01
 #define DEFAULT_CPU_INDEX_TUPLE_COST 0.001
 #define DEFAULT_CPU_OPERATOR_COST  0.0025
 
+#define DEFAULT_EFFECTIVE_CACHE_SIZE  1000.0	/* measured in pages */
+
 
 /*
  * prototypes for costsize.c
@@ -34,11 +36,12 @@
  */
 
 /* parameter variables and flags */
-extern double effective_cache_size;
-extern double random_page_cost;
-extern double cpu_tuple_cost;
+extern DLLIMPORT double seq_page_cost;
+extern DLLIMPORT double random_page_cost;
+extern DLLIMPORT double cpu_tuple_cost;
 extern DLLIMPORT double cpu_index_tuple_cost;
-extern double cpu_operator_cost;
+extern DLLIMPORT double cpu_operator_cost;
+extern double effective_cache_size;
 extern Cost disable_cost;
 extern bool enable_seqscan;
 extern bool enable_indexscan;