Add FILLFACTOR to CREATE INDEX.

ITAGAKI Takahiro

src/backend/access/nbtree/nbtinsert.c

@@ -8,13 +8,14 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/access/nbtree/nbtinsert.c,v 1.137 2006/05/08 00:00:09 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/access/nbtree/nbtinsert.c,v 1.138 2006/07/02 02:23:18 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
 
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/heapam.h"
 #include "access/nbtree.h"
 #include "miscadmin.h"
@@ -25,6 +26,7 @@ typedef struct
 {
 	/* context data for _bt_checksplitloc */
 	Size		newitemsz;		/* size of new item to be inserted */
+	int			fillfactor;		/* used when insert at right most */
 	bool		is_leaf;		/* T if splitting a leaf page */
 	bool		is_rightmost;	/* T if splitting a rightmost page */
 
@@ -986,14 +988,11 @@ _bt_split(Relation rel, Buffer buf, OffsetNumber firstright,
  * it needs to go into!)
  *
  * If the page is the rightmost page on its level, we instead try to arrange
- * for twice as much free space on the right as on the left.  In this way,
+ * for reserving (100-fillfactor)% of free space on left page. In this way,
  * when we are inserting successively increasing keys (consider sequences,
- * timestamps, etc) we will end up with a tree whose pages are about 67% full,
+ * timestamps, etc) we will end up with a tree whose pages are about fillfactor% full,
  * instead of the 50% full result that we'd get without this special case.
- * (We could bias it even further to make the initially-loaded tree more full.
- * But since the steady-state load for a btree is about 70%, we'd likely just
- * be making more page-splitting work for ourselves later on, when we start
- * seeing updates to existing tuples.)
+ * This is the same as initially-loaded tree.
  *
  * We are passed the intended insert position of the new tuple, expressed as
  * the offsetnumber of the tuple it must go in front of.  (This could be
@@ -1027,6 +1026,7 @@ _bt_findsplitloc(Relation rel,
 	/* Passed-in newitemsz is MAXALIGNED but does not include line pointer */
 	newitemsz += sizeof(ItemIdData);
 	state.newitemsz = newitemsz;
+	state.fillfactor = IndexGetFillFactor(rel);
 	state.is_leaf = P_ISLEAF(opaque);
 	state.is_rightmost = P_RIGHTMOST(opaque);
 	state.have_split = false;
@@ -1157,10 +1157,11 @@ _bt_checksplitloc(FindSplitData *state, OffsetNumber firstright,
 		if (state->is_rightmost)
 		{
 			/*
-			 * On a rightmost page, try to equalize right free space with
-			 * twice the left free space.  See comments for _bt_findsplitloc.
+			 * On a rightmost page, try to reserve (100-fillfactor)% of
+			 * free space on left page. See comments for _bt_findsplitloc.
 			 */
-			delta = (2 * leftfree) - rightfree;
+			delta = (state->fillfactor * leftfree)
+				- ((100 - state->fillfactor) * rightfree);
 		}
 		else
 		{

src/backend/access/nbtree/nbtsort.c

@@ -56,13 +56,14 @@
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/access/nbtree/nbtsort.c,v 1.102 2006/06/27 16:53:02 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/access/nbtree/nbtsort.c,v 1.103 2006/07/02 02:23:19 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
 
 #include "postgres.h"
 
+#include "access/genam.h"
 #include "access/nbtree.h"
 #include "access/xlog.h"
 #include "miscadmin.h"
@@ -120,6 +121,7 @@ typedef struct BTWriteState
 
 
 static Page _bt_blnewpage(uint32 level);
+static Size	_bt_full_threshold(Relation index, Size pagesize, bool leaf);
 static BTPageState *_bt_pagestate(BTWriteState *wstate, uint32 level);
 static void _bt_slideleft(Page page);
 static void _bt_sortaddtup(Page page, Size itemsize,
@@ -327,6 +329,22 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
 	pfree(page);
 }
 
+/*
+ * The steady-state load factor for btrees is usually estimated at 70%.
+ * We choose to pack leaf pages to 90% and upper pages to 70% as defaults.
+ */
+static Size
+_bt_full_threshold(Relation index, Size pagesize, bool leaf)
+{
+	int fillfactor = IndexGetFillFactor(index);
+	if (!leaf)
+	{
+		/* XXX: Is this reasonable? */
+		fillfactor = Max(70, 3 * fillfactor - 200);
+	}
+	return pagesize * (100 - fillfactor) / 100;
+}
+
 /*
  * allocate and initialize a new BTPageState.  the returned structure
  * is suitable for immediate use by _bt_buildadd.
@@ -347,10 +365,8 @@ _bt_pagestate(BTWriteState *wstate, uint32 level)
 	state->btps_lastoff = P_HIKEY;
 	state->btps_level = level;
 	/* set "full" threshold based on level.  See notes at head of file. */
-	if (level > 0)
-		state->btps_full = (PageGetPageSize(state->btps_page) * 3) / 10;
-	else
-		state->btps_full = PageGetPageSize(state->btps_page) / 10;
+	state->btps_full = _bt_full_threshold(wstate->index,
+		PageGetPageSize(state->btps_page), level == 0);
 	/* no parent level, yet */
 	state->btps_next = NULL;
 

src/backend/access/nbtree/nbtutils.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/access/nbtree/nbtutils.c,v 1.74 2006/05/08 00:00:10 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/access/nbtree/nbtutils.c,v 1.75 2006/07/02 02:23:19 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -1079,3 +1079,16 @@ BTreeShmemInit(void)
 	else
 		Assert(found);
 }
+
+Datum
+btoption(PG_FUNCTION_ARGS)
+{
+#define BTREE_MIN_FILLFACTOR		50
+#define BTREE_DEFAULT_FILLFACTOR	90
+
+	ArrayType	   *options = (ArrayType *) PG_GETARG_POINTER(0);
+
+	/* Use index common routine. */
+	PG_RETURN_BYTEA_P(genam_option(options,
+		BTREE_MIN_FILLFACTOR, BTREE_DEFAULT_FILLFACTOR));
+}