我尝试运行2个查询，区别仅在于我将状态从“CA”更改为“FL”的where子句。并且执行时间增加了60倍你可以用正确的方式指出我，为什么两个符号会如此改变查询时间？
第一次查询：
SELECT (COUNT(DISTINCT "Location"."doctor_id")) AS "Location__td",
(COUNT(DISTINCT "Location"."id")) AS "Location__tl"
FROM "public"."locations" AS "Location"
JOIN
(SELECT location_id, (array_agg("DoctorP"."fullname"))[1] fullname, (array_agg("DoctorP"."license"))[1] license, (array_agg("DoctorP"."gender"))[1] gender
FROM services
JOIN
(SELECT id,
fullname,
license,
gender
FROM "public"."doctors" AS "DoctorP"
WHERE "DoctorP"."id" NOT IN
(SELECT doctor_id
FROM folder_doctors
WHERE folder_id IN (26,
25,
24))) AS "DoctorP" ON "DoctorP"."id" = doctor_id
WHERE plan_desc = 'Commercial'
GROUP BY location_id) AS "Service" ON ("Location"."id" = "Service"."location_id")
WHERE coordinates IS NOT NULL
AND ST_Contains(ST_MakeEnvelope(-57.121123689709975, 50.33132725216629, -128.31252993970998, 18.208317875535407), "Location"."coordinates")
AND "Location"."state" = 'CA'
LIMIT 1
22478
1 строка (0.420 s)
第二次查询：
SELECT (COUNT(DISTINCT "Location"."doctor_id")) AS "Location__td",
(COUNT(DISTINCT "Location"."id")) AS "Location__tl"
FROM "public"."locations" AS "Location"
JOIN
(SELECT location_id, (array_agg("DoctorP"."fullname"))[1] fullname, (array_agg("DoctorP"."license"))[1] license, (array_agg("DoctorP"."gender"))[1] gender
FROM services
JOIN
(SELECT id,
fullname,
license,
gender
FROM "public"."doctors" AS "DoctorP"
WHERE "DoctorP"."id" NOT IN
(SELECT doctor_id
FROM folder_doctors
WHERE folder_id IN (26,
25,
24))) AS "DoctorP" ON "DoctorP"."id" = doctor_id
WHERE plan_desc = 'Commercial'
GROUP BY location_id) AS "Service" ON ("Location"."id" = "Service"."location_id")
WHERE coordinates IS NOT NULL
AND ST_Contains(ST_MakeEnvelope(-57.121123689709975, 50.33132725216629, -128.31252993970998, 18.208317875535407), "Location"."coordinates")
AND "Location"."state" = 'FL'
LIMIT 1
399
1 строка (27.734 s)
第一个查询的说明：
Limit (cost=36592.78..36592.79 rows=1 width=16)
-> Aggregate (cost=36592.78..36592.79 rows=1 width=16)
-> Hash Join (cost=36303.68..36592.76 rows=4 width=16)
Hash Cond: (services.location_id = "Location".id)
-> HashAggregate (cost=36059.78..36181.48 rows=12170 width=8)
Group Key: services.location_id
-> Hash Join (cost=1353.90..35573.63 rows=194461 width=8)
Hash Cond: (services.doctor_id = "DoctorP".id)
-> Seq Scan on services (cost=0.00..30816.61 rows=388935 width=16)
Filter: (plan_desc = 'Commercial'::text)
-> Hash (cost=1154.22..1154.22 rows=15974 width=8)
-> Seq Scan on doctors "DoctorP" (cost=20.86..1154.22 rows=15974 width=8)
Filter: (NOT (hashed SubPlan 1))
SubPlan 1
-> Index Scan using folder_doctors_folder_id_index on folder_doctors (cost=0.28..20.78 rows=34 width=8)
Index Cond: (folder_id = ANY ('{26,25,24}'::integer[]))
-> Hash (cost=243.66..243.66 rows=19 width=16)
-> Bitmap Heap Scan on locations "Location" (cost=4.88..243.66 rows=19 width=16)
Recheck Cond: ((coordinates IS NOT NULL) AND ('01030000000100000005000000A00927FB808F4CC0F42970EE682A4940A00927FB808F4CC0C698FE515435324068C2C93E000A60C0C698FE515435324068C2C93E000A60C0F42970EE682A4940A00927FB808F4CC0F42970EE682A4940'::geometry ~ coordinates))
Filter: ((state = 'CA'::bpchar) AND _st_contains('01030000000100000005000000A00927FB808F4CC0F42970EE682A4940A00927FB808F4CC0C698FE515435324068C2C93E000A60C0C698FE515435324068C2C93E000A60C0F42970EE682A4940A00927FB808F4CC0F42970EE682A4940'::geometry, coordinates))
-> Bitmap Index Scan on locations_coordinates_gix (cost=0.00..4.88 rows=60 width=0)
Index Cond: ((coordinates IS NOT NULL) AND ('01030000000100000005000000A00927FB808F4CC0F42970EE682A4940A00927FB808F4CC0C698FE515435324068C2C93E000A60C0C698FE515435324068C2C93E000A60C0F42970EE682A4940A00927FB808F4CC0F42970EE682A4940'::geometry ~ coordinates))
第二个查询的解释：
Limit (cost=36550.09..36550.10 rows=1 width=16)
-> Aggregate (cost=36550.09..36550.10 rows=1 width=16)
-> Nested Loop (cost=36137.80..36550.09 rows=1 width=16)
Join Filter: ("Location".id = services.location_id)
-> Bitmap Heap Scan on locations "Location" (cost=78.02..94.78 rows=1 width=16)
Recheck Cond: ((coordinates IS NOT NULL) AND ('01030000000100000005000000A00927FB808F4CC0F42970EE682A4940A00927FB808F4CC0C698FE515435324068C2C93E000A60C0C698FE515435324068C2C93E000A60C0F42970EE682A4940A00927FB808F4CC0F42970EE682A4940'::geometry ~ coordinates) AND (state = 'FL'::bpchar))
Filter: _st_contains('01030000000100000005000000A00927FB808F4CC0F42970EE682A4940A00927FB808F4CC0C698FE515435324068C2C93E000A60C0C698FE515435324068C2C93E000A60C0F42970EE682A4940A00927FB808F4CC0F42970EE682A4940'::geometry, coordinates)
-> BitmapAnd (cost=78.02..78.02 rows=4 width=0)
-> Bitmap Index Scan on locations_coordinates_gix (cost=0.00..4.88 rows=60 width=0)
Index Cond: ((coordinates IS NOT NULL) AND ('01030000000100000005000000A00927FB808F4CC0F42970EE682A4940A00927FB808F4CC0C698FE515435324068C2C93E000A60C0C698FE515435324068C2C93E000A60C0F42970EE682A4940A00927FB808F4CC0F42970EE682A4940'::geometry ~ coordinates))
-> Bitmap Index Scan on locations_state (cost=0.00..72.89 rows=3813 width=0)
Index Cond: (state = 'FL'::bpchar)
-> HashAggregate (cost=36059.78..36181.48 rows=12170 width=8)
Group Key: services.location_id
-> Hash Join (cost=1353.90..35573.63 rows=194461 width=8)
Hash Cond: (services.doctor_id = "DoctorP".id)
-> Seq Scan on services (cost=0.00..30816.61 rows=388935 width=16)
Filter: (plan_desc = 'Commercial'::text)
-> Hash (cost=1154.22..1154.22 rows=15974 width=8)
-> Seq Scan on doctors "DoctorP" (cost=20.86..1154.22 rows=15974 width=8)
Filter: (NOT (hashed SubPlan 1))
SubPlan 1
-> Index Scan using folder_doctors_folder_id_index on folder_doctors (cost=0.28..20.78 rows=34 width=8)
Index Cond: (folder_id = ANY ('{26,25,24}'::integer[]))