Question

我有一个Postgis查询，从位置选择给定半径内的所有点。

SELECT id 
FROM mytable 
WHERE ST_DWithin(location, ST_GeographyFromText('SRID=4326;POINT(12.1234 12.1234)'), 789) 
OR ST_DWithin(location, ST_GeographyFromText('SRID=4326;POINT(23.4567 23.4567)'), 789) 
OR ST_DWithin(location, ST_GeographyFromText('SRID=4326;POINT(34.5678 34.5678)'), 789)

当我用EXPLAIN分析时，我可以看到

Bitmap Heap Scan on mytable  (cost=964.16..1051.01 rows=1 width=4)
Recheck Cond: ((location && 'E61A301BC000002OBFUSCATEDDAT05123240'::geography) OR (location && '010020E610OBFUSCATEDDATAC9ACE4E0E3240'::geography) OR (location && '0101000OBFUSCATEDDATA8FC3042080020E613240'::geography) )
Filter: (((location && 'E61A301BC000002OBFUSCATEDDAT05123240'::geography) AND ('010020E610OBFUSCATEDDATAC9ACE4E0E3240'::geography && _st_e...SOMETHING
->  BitmapOr  (cost=964.16..964.16 rows=1 width=0)
  ->  Bitmap Index Scan on global_points_gix  (cost=0.00..4.42 rows=1 width=0)
  Index Cond: (location && 'E61A301BC000002OBFUSCATEDDAT05123240'::geography)
  ->  Bitmap Index Scan on global_points_gix  (cost=0.00..4.42 rows=1 width=0)
  Index Cond: (location && '010020E610OBFUSCATEDDATAC9ACE4E0E3240'::geography)
  ->  Bitmap Index Scan on global_points_gix  (cost=0.00..4.42 rows=1 width=0)
  Index Cond: (location && '0101000020OBFUSCATEDDATA8FC3042080020E613240'::geography)

这让我觉得搜索积分非常快，但合并结果排除重复是很昂贵的。我对吗？如果我对它们没问题，我能以某种方式启用重复吗？我尝试将OR替换为UNION SELECT，但它没有成功。

在Postgis查询中启用重复项以加快查询速度

0 个答案: