@@ -2758,7 +2758,7 @@ include 'filename'
2758
2758
2759
2759
<para>
2760
2760
Random access to mechanical disk storage is normally much more expensive
2761
- than four- times sequential access. However, a lower default is used
2761
+ than four times sequential access. However, a lower default is used
2762
2762
(4.0) because the majority of random accesses to disk, such as indexed
2763
2763
reads, are assumed to be in cache. The default value can be thought of
2764
2764
as modeling random access as 40 times slower than sequential, while
@@ -2841,9 +2841,17 @@ include 'filename'
2841
2841
<listitem>
2842
2842
<para>
2843
2843
Sets the planner's assumption about the effective size of the
2844
- disk cache that is available to a single query. This is
2845
- factored into estimates of the cost of using an index; a
2846
- higher value makes it more likely index scans will be used, a
2844
+ disk cache that is available to a single query. The default
2845
+ setting of -1 selects a size equal to four times the size of <xref
2846
+ linkend="guc-shared-buffers">, but not less than the size of one
2847
+ shared buffer page, typically <literal>8kB</literal>. This value
2848
+ can be set manually if the automatic choice is too large or too
2849
+ small.
2850
+ </para>
2851
+
2852
+ <para>
2853
+ This value is factored into estimates of the cost of using an index;
2854
+ a higher value makes it more likely index scans will be used, a
2847
2855
lower value makes it more likely sequential scans will be
2848
2856
used. When setting this parameter you should consider both
2849
2857
<productname>PostgreSQL</productname>'s shared buffers and the
@@ -2855,8 +2863,10 @@ include 'filename'
2855
2863
memory allocated by <productname>PostgreSQL</productname>, nor
2856
2864
does it reserve kernel disk cache; it is used only for estimation
2857
2865
purposes. The system also does not assume data remains in
2858
- the disk cache between queries. The default is 128 megabytes
2859
- (<literal>128MB</>).
2866
+ the disk cache between queries. The auto-tuning
2867
+ selected by the default setting of -1 should give reasonable
2868
+ results if this database cluster is can utilize most of the memory
2869
+ on this server.
2860
2870
</para>
2861
2871
</listitem>
2862
2872
</varlistentry>
0 commit comments