我有7个表加入单选查询。但是，当我加入5个表中的任何一个时，它工作正常但加入7个表会使Query无响应。而只有大约2k条记录可供选择。这是我的疑问：
select c.i_cdr, v.i_cdrs_connection, c.i_call,
cs.name AS customer_name, a.username AS client_name_id,
con.name AS connection_name, vn.name AS vendor_name_id,
ca.setup_time,
c.remote_ip AS c_remote_ip_id, con.destination AS v_remote_ip_id
FROM cdrs c
JOIN Calls ca using (i_call)
JOIN cdrs_connections v using (i_call)
JOIN accounts a using(i_account)
JOIN customers cs using (i_customer)
JOIN connections con using(i_connection)
JOIN vendors vn using (i_vendor)
where i_cdr>397233018
如果单个查询中的最大表数或最大列数有限制吗？
我尝试过减少列但没有用。
postgresql配置。
postgres 8.4
shared_buffers = 128MB
work_mem = 32MB
maintenance_work_mem = 64MB
调用表：
Create Table calls
(
i_call serial Not null,
setup_time timestamp Not Null,
cli character varying(64),
cld character varying(64)
)
cdrs表：
Create Table cdrs
(
i_cdr serial Not null,
i_call bigint Not Null,
remote_ip character varying(64),
i_account bigint Not Null
//few more not used fields
)
帐户表：
Create Table accounts
(
i_account serial Not null,
i_customer bigint Not Null,
username character varying(128)
)
客户表：
Create Table customers
(
i_customer serial Not null,
name character varying(128)
)
cdrs_connections：
Create Table cdrs_connections
(
i_cdrs_connection serial Not null,
i_call bigint Not null,
i_connection bigint Not null,
connect_time timestamp,
cld_out character varying(64),
cli_out character varying(128)
)
连接：
Create Table connections
(
i_connection serial Not null,
i_vendor bigint Not null,
name varying character(64),
destination character varying(64)
)
供应商：
Create Table vendors
(
i_vendor serial Not null,
name character varying(64)
);
解释追踪：
遍历的主要表被分区为1，即使我排除有500行的帐户和有26行的客户也可以正常工作，这样很好。
Nested Loop (cost=3388.21..3202797988160500.00 rows=6369283937505514496 width=122)
Join Filter: (c.i_call = ca.i_call)
-> Nested Loop (cost=3388.21..29281315906299.16 rows=12704473261254 width=122)
-> Hash Join (cost=3388.21..310816005731.24 rows=12704473261254 width=121)
Hash Cond: (c.i_account = a.i_account)
-> Nested Loop (cost=0.00..24965353964.82 rows=12704473261254 width=112)
Join Filter: (v.i_call = c.i_call)
-> Nested Loop (cost=0.00..1708130302.64 rows=61256189 width=69)
-> Nested Loop (cost=0.00..465977965.21 rows=61256189 width=60)
-> Append (cost=0.00..6600178.80 rows=61256189 width=24)
-> Index Scan using cdrs_connections_i_call on cdrs_connections v (cost=0.00..22.03 rows=310 width=24)
-> Index Scan using cdrs_connections1_i_call on cdrs_connections1 v (cost=0.00..219415.43 rows=568528 width=24)
-> Index Scan using cdrs_connections2_i_call on cdrs_connections2 v (cost=0.00..34433.40 rows=536331 width=24)
-> Index Scan using cdrs_connections3_i_call on cdrs_connections3 v (cost=0.00..359871.77 rows=885115 width=24)
-> Index Scan using cdrs_connections4_i_call on cdrs_connections4 v (cost=0.00..84725.16 rows=1316695 width=24)
-> Index Scan using cdrs_connections5_i_call on cdrs_connections5 v (cost=0.00..66291.88 rows=1046102 width=24)
-> Index Scan using cdrs_connections6_i_call on cdrs_connections6 v (cost=0.00..74410.59 rows=1043184 width=24)
-> Index Scan using cdrs_connections7_i_call on cdrs_connections7 v (cost=0.00..79614.35 rows=1244914 width=24)
-> Index Scan using cdrs_connections8_i_call on cdrs_connections8 v (cost=0.00..78292.13 rows=1228840 width=24)
-> Index Scan using cdrs_connections9_i_call on cdrs_connections9 v (cost=0.00..23.84 rows=320 width=24)
-> Index Scan using cdrs_connections10_i_call on cdrs_connections10 v (cost=0.00..869491.71 rows=1326456 width=24)
-> Index Scan using cdrs_connections11_i_call on cdrs_connections11 v (cost=0.00..84478.96 rows=1329600 width=24)
-> Index Scan using cdrs_connections12_i_call on cdrs_connections12 v (cost=0.00..34373.13 rows=529498 width=24)
-> Index Scan using cdrs_connections13_i_call on cdrs_connections13 v (cost=0.00..47.28 rows=330 width=24)
-> Index Scan using cdrs_connections14_i_call on cdrs_connections14 v (cost=0.00..33623.46 rows=506174 width=24)
-> Index Scan using cdrs_connections15_i_call on cdrs_connections15 v (cost=0.00..31732.57 rows=485078 width=24)
-> Index Scan using cdrs_connections16_i_call on cdrs_connections16 v (cost=0.00..29434.18 rows=460046 width=24)
-> Index Scan using cdrs_connections17_i_call on cdrs_connections17 v (cost=0.00..32157.93 rows=500074 width=24)
-> Index Scan using cdrs_connections18_i_call on cdrs_connections18 v (cost=0.00..35811.81 rows=561528 width=24)
-> Index Scan using cdrs_connections19_i_call on cdrs_connections19 v (cost=0.00..157742.69 rows=835540 width=24)
-> Index Scan using cdrs_connections20_i_call on cdrs_connections20 v (cost=0.00..104175.88 rows=1638302 width=24)
-> Index Scan using cdrs_connections21_i_call on cdrs_connections21 v (cost=0.00..232173.00 rows=1859055 width=24)
-> Index Scan using cdrs_connections22_i_call on cdrs_connections22 v (cost=0.00..99680.31 rows=1549844 width=24)
-> Index Scan using cdrs_connections23_i_call on cdrs_connections23 v (cost=0.00..913103.99 rows=1597071 width=24)
-> Index Scan using cdrs_connections24_i_call on cdrs_connections24 v (cost=0.00..151771.48 rows=1547203 width=24)
-> Index Scan using cdrs_connections25_i_call on cdrs_connections25 v (cost=0.00..48.24 rows=320 width=24)
-> Index Scan using cdrs_connections26_i_call on cdrs_connections26 v (cost=0.00..94067.11 rows=1472598 width=24)
-> Index Scan using cdrs_connections27_i_call on cdrs_connections27 v (cost=0.00..110525.50 rows=1083587 width=24)
-> Index Scan using cdrs_connections28_i_call on cdrs_connections28 v (cost=0.00..352381.02 rows=1068230 width=24)
-> Index Scan using cdrs_connections29_i_call on cdrs_connections29 v (cost=0.00..78430.93 rows=1234409 width=24)
-> Index Scan using cdrs_connections30_i_call on cdrs_connections30 v (cost=0.00..44659.37 rows=696125 width=24)
-> Index Scan using connections_pkey on connections con (cost=0.00..7.49 rows=1 width=52)
Index Cond: (con.i_connection = v.i_connection)
-> Index Scan using vendors_pkey on vendors vn (cost=0.00..20.27 rows=1 width=25)
Index Cond: (vn.i_vendor = con.i_vendor)
-> Append (cost=0.00..378.91 rows=61 width=43)
-> Index Scan using cdrs_i_call on cdrs c (cost=0.00..3.87 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs1_i_call on cdrs1 c (cost=0.00..8.32 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs2_i_call on cdrs2 c (cost=0.00..8.32 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs3_i_call on cdrs3 c (cost=0.00..8.38 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs4_i_call on cdrs4 c (cost=0.00..8.45 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs5_i_call on cdrs5 c (cost=0.00..8.42 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs6_i_call on cdrs6 c (cost=0.00..8.43 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs7_i_call on cdrs7 c (cost=0.00..8.48 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs8_i_call on cdrs8 c (cost=0.00..8.44 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs9_i_call on cdrs9 c (cost=0.00..2.27 rows=1 width=38)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs10_i_call on cdrs10 c (cost=0.00..8.45 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs11_i_call on cdrs11 c (cost=0.00..8.44 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs12_i_call on cdrs12 c (cost=0.00..8.32 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs13_i_call on cdrs13 c (cost=0.00..3.87 rows=1 width=38)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs14_i_call on cdrs14 c (cost=0.00..8.32 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs15_i_call on cdrs15 c (cost=0.00..8.31 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs16_i_call on cdrs16 c (cost=0.00..8.31 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs17_i_call on cdrs17 c (cost=0.00..8.32 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs18_i_call on cdrs18 c (cost=0.00..8.32 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs19_i_call on cdrs19 c (cost=0.00..8.36 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs20_i_call on cdrs20 c (cost=0.00..8.48 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs21_i_call on cdrs21 c (cost=0.00..8.51 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs22_i_call on cdrs22 c (cost=0.00..8.47 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs23_i_call on cdrs23 c (cost=0.00..8.47 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs24_i_call on cdrs24 c (cost=0.00..8.47 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs25_i_call on cdrs25 c (cost=0.00..3.87 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs26_i_call on cdrs26 c (cost=0.00..8.49 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs27_i_call on cdrs27 c (cost=0.00..8.42 rows=1 width=37)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs28_i_call on cdrs28 c (cost=0.00..8.41 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs29_i_call on cdrs29 c (cost=0.00..8.43 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Index Scan using cdrs30_i_call on cdrs30 c (cost=0.00..8.35 rows=1 width=36)
Index Cond: (c.i_call = v.i_call)
Filter: (c.i_cdr > 39726118)
-> Hash (cost=3001.41..3001.41 rows=30944 width=25)
-> Index Scan using accounts_i_customer on accounts a (cost=0.00..3001.41 rows=30944 width=25)
-> Index Scan using customers_pkey on customers cs (cost=0.00..2.27 rows=1 width=17)
Index Cond: (cs.i_customer = a.i_customer)
-> Append (cost=0.00..249.41 rows=31 width=16)
-> Index Scan using calls_pkey on calls ca (cost=0.00..3.87 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls1_pkey on calls1 ca (cost=0.00..8.42 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls2_pkey on calls2 ca (cost=0.00..8.42 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls3_pkey on calls3 ca (cost=0.00..8.63 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls4_pkey on calls4 ca (cost=0.00..9.07 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls5_pkey on calls5 ca (cost=0.00..8.84 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls6_pkey on calls6 ca (cost=0.00..8.90 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls7_pkey on calls7 ca (cost=0.00..8.95 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls8_pkey on calls8 ca (cost=0.00..8.65 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls9_pkey on calls9 ca (cost=0.00..3.87 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls10_pkey on calls10 ca (cost=0.00..8.78 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls11_pkey on calls11 ca (cost=0.00..8.75 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls12_pkey on calls12 ca (cost=0.00..8.40 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls13_pkey on calls13 ca (cost=0.00..3.87 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls14_pkey on calls14 ca (cost=0.00..8.40 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls15_pkey on calls15 ca (cost=0.00..8.42 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls16_pkey on calls16 ca (cost=0.00..8.39 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls17_pkey on calls17 ca (cost=0.00..8.40 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls18_pkey on calls18 ca (cost=0.00..8.40 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls19_pkey on calls19 ca (cost=0.00..8.53 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls20_pkey on calls20 ca (cost=0.00..8.78 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls21_pkey on calls21 ca (cost=0.00..8.84 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls22_pkey on calls22 ca (cost=0.00..8.78 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls23_pkey on calls23 ca (cost=0.00..8.79 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls24_pkey on calls24 ca (cost=0.00..8.74 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls25_pkey on calls25 ca (cost=0.00..3.87 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls26_pkey on calls26 ca (cost=0.00..9.45 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls27_pkey on calls27 ca (cost=0.00..8.56 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls28_pkey on calls28 ca (cost=0.00..8.60 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls29_pkey on calls29 ca (cost=0.00..8.60 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
-> Index Scan using calls30_pkey on calls30 ca (cost=0.00..8.44 rows=1 width=16)
Index Cond: (ca.i_call = c.i_call)
(326 rows)
提示：
我不确切知道为什么但是当我使用给定查询创建View而没有最后两个表连接和供应商然后将此View与其余表连接查询工作正常。
解决方案： 我通过改变连接顺序解决了这个问题。 从：
select c.i_cdr, v.i_cdrs_connection, c.i_call,
cs.name AS customer_name, a.username AS client_name_id,
con.name AS connection_name, vn.name AS vendor_name_id,
ca.setup_time,
c.remote_ip AS c_remote_ip_id, con.destination AS v_remote_ip_id
FROM cdrs c
JOIN Calls ca using (i_call)
JOIN cdrs_connections v using (i_call)
JOIN accounts a using(i_account)
JOIN customers cs using (i_customer)
JOIN connections con using(i_connection)
JOIN vendors vn using (i_vendor)
where i_cdr>397233018
要：
select c.i_cdr, v.i_cdrs_connection, c.i_call,
cs.name AS customer_name, a.username AS client_name_id,
con.name AS connection_name, vn.name AS vendor_name_id,
ca.setup_time,
c.remote_ip AS c_remote_ip_id, con.destination AS v_remote_ip_id
FROM cdrs c
JOIN accounts a using(i_account)
JOIN customers cs using (i_customer)
JOIN Calls ca using (i_call)
JOIN cdrs_connections v using (i_call)
JOIN connections con using(i_connection)
JOIN vendors vn using (i_vendor)
where i_cdr>397233018
这是因为cdrs表链接到accounts表，而account表又连接到customers表。供应商表通过连接表连接到cdrs_connections。因此，通过放置这些表按顺序删除循环。现在响应时间非常好。
答案 0 :(得分：2)
有关最大列的信息，请参阅the Limits heading in About PostgreSQL。
连接中是否有最大数量的表我不确定。在实践中，真正的限制将是时间和计算机的性能以及RAM +磁盘空间。
您确实需要使用EXPLAIN ANALYZE检查您获得的查询计划，以了解缓慢的原因和原因。 explain.depesz.com可以帮助提高结果的可读性。
注意确保您没有收到任何不受欢迎的cartesian product。但是，您可能没有使用INNER JOIN ... USING语法。
答案 1 :(得分：0)
只需在查询中添加“按i_cdr排序”
即可