場(chǎng)景 lower(name) like 'pf%'
create table users (id int primary key, name varchar(255));
Create or replace function random_string(length integer) returns text as
$$
declare
chars text[] := '{0,1,2,3,4,5,6,7,8,9,A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V,W,X,Y,Z,a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z}';
result text := '';
i integer := 0;
begin
if length 0 then
raise exception 'Given length cannot be less than 0';
end if;
for i in 1..length loop
result := result || chars[1+random()*(array_length(chars, 1)-1)];
end loop;
return result;
end;
$$ language plpgsql;
insert into users values(generate_series(1,50000), random_string(15));
普通bt：不走索引
pg_trgm模塊提供函數(shù)和操作符測(cè)定字母數(shù)字文本基于三元模型匹配的相似性，還有支持快速搜索相似字符串的索引操作符類。三元模型是一組從一個(gè)字符串中獲得的三個(gè)連續(xù)的字符。我們可以通過計(jì)數(shù)兩個(gè)字符串共享的三元模型的數(shù)量來測(cè)量它們的相似性。這個(gè)簡(jiǎn)單的想法證明在測(cè)量許多自然語(yǔ)言詞匯的相似性時(shí)是非常有效的。
CREATE INDEX users_idx0 ON users (name);
全字匹配查詢（走索引）
explain select * from users where name='pfDNQVmhqDrF1EY';
QUERY PLAN
-------------------------------------------------------------------------
Index Scan using users_idx0 on users (cost=0.29..8.31 rows=1 width=20)
Index Cond: ((name)::text = 'pfDNQVmhqDrF1EY'::text)
(2 rows)
加函數(shù)全字匹配（不走索引）
explain select * from users where lower(name)='pfDNQVmhqDrF1EY';
QUERY PLAN
-----------------------------------------------------------
Seq Scan on users (cost=0.00..1069.00 rows=250 width=20)
Filter: (lower((name)::text) = 'pfDNQVmhqDrF1EY'::text)
(2 rows)
模糊匹配（不走索引）
explain select * from users where name like 'pf%';
QUERY PLAN
--------------------------------------------------------
Seq Scan on users (cost=0.00..944.00 rows=5 width=20)
Filter: ((name)::text ~~ 'pf%'::text)
explain select * from users where name like 'pf_';
QUERY PLAN
--------------------------------------------------------
Seq Scan on users (cost=0.00..944.00 rows=5 width=20)
Filter: ((name)::text ~~ 'pf_'::text)
字段帶函數(shù)的bt索引：函數(shù)走索引
drop index users_idx0;
CREATE INDEX users_dex1 ON users (lower(name));
加函數(shù)全字匹配（走索引）
explain select * from users where lower(name)='pfDNQVmhqDrF1EY';
QUERY PLAN
---------------------------------------------------------------------------
Bitmap Heap Scan on users (cost=6.23..324.34 rows=250 width=20)
Recheck Cond: (lower((name)::text) = 'pfDNQVmhqDrF1EY'::text)
-> Bitmap Index Scan on users_dex1 (cost=0.00..6.17 rows=250 width=0)
Index Cond: (lower((name)::text) = 'pfDNQVmhqDrF1EY'::text)
(4 rows)
模糊匹配（不走索引）
explain select * from users where lower(name) like 'pf%';
QUERY PLAN
-----------------------------------------------------------
Seq Scan on users (cost=0.00..1069.00 rows=250 width=20)
Filter: (lower((name)::text) ~~ 'pf%'::text)
(2 rows)
聲明操作符類的bt索引：like走索引
定義索引的同時(shí)可以為索引的每個(gè)字段聲明一個(gè)操作符類。
CREATE INDEX name ON table (column opclass [sort options] [, …]);
這個(gè)操作符類指明該索引用于該字段時(shí)要使用的操作符。
CREATE INDEX users_dex2 ON users (lower(name) varchar_pattern_ops);
模糊匹配（走索引）
explain select * from users where lower(name) like 'pf%';
QUERY PLAN
------------------------------------------------------------------------------------------------------
Bitmap Heap Scan on users (cost=4.82..144.00 rows=5 width=20)
Filter: (lower((name)::text) ~~ 'pf%'::text)
-> Bitmap Index Scan on users_dex2 (cost=0.00..4.82 rows=53 width=0)
Index Cond: ((lower((name)::text) ~>=~ 'pf'::text) AND (lower((name)::text) ~~ 'pg'::text))
(4 rows)
場(chǎng)景2 name like '%pf%'
Create or replace function random_string(length integer) returns text as
$$
declare
chars text[] := '{0,1,2,3,4,5,6,7,8,9,A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V,W,X,Y,Z,a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z}';
result text := '';
i integer := 0;
begin
if length 0 then
raise exception 'Given length cannot be less than 0';
end if;
for i in 1..length loop
result := result || chars[1+random()*(array_length(chars, 1)-1)];
end loop;
return result;
end;
$$ language plpgsql;
create table users (id int primary key, name varchar(255));
insert into users values(generate_series(1,50000), random_string(15));
聲明操作符bt：不走索引
CREATE INDEX idx_name ON users USING btree (lower(name) varchar_pattern_ops);
explain (analyze true,format yaml, verbose true, buffers true) select * from users where lower(name) like '%pf%';\
QUERY PLAN
-----------------------------------------------------------
- Plan: +
Node Type: "Seq Scan" +
Parallel Aware: false +
Relation Name: "users" +
Schema: "public" +
Alias: "users" +
Startup Cost: 0.00 +
Total Cost: 1069.00 +
Plan Rows: 5 +
Plan Width: 20 +
Actual Startup Time: 0.320 +
Actual Total Time: 86.841 +
Actual Rows: 710 +
Actual Loops: 1 +
Output: +
- "id" +
- "name" +
Filter: "(lower((users.name)::text) ~~ '%pf%'::text)"+
Rows Removed by Filter: 49290 +
Shared Hit Blocks: 319 +
Shared Read Blocks: 0 +
Shared Dirtied Blocks: 0 +
Shared Written Blocks: 0 +
Local Hit Blocks: 0 +
Local Read Blocks: 0 +
Local Dirtied Blocks: 0 +
Local Written Blocks: 0 +
Temp Read Blocks: 0 +
Temp Written Blocks: 0 +
Planning Time: 0.188 +
Triggers: +
Execution Time: 86.975
聲明pg_trgm操作符bt：可以走索引
CREATE EXTENSION pg_trgm;
CREATE INDEX idx_users_name_trgm_gist ON users USING gist (name gist_trgm_ops);
explain (analyze true, verbose true, buffers true) select * from users where name like '%pf%';
QUERY PLAN
------------------------------------------------------------------------------------------------------------------------------------------
Bitmap Heap Scan on public.users (cost=32.19..371.08 rows=505 width=20) (actual time=19.314..53.132 rows=193 loops=1)
Output: id, name
Recheck Cond: ((users.name)::text ~~ '%pf%'::text)
Rows Removed by Index Recheck: 49807
Heap Blocks: exact=319
Buffers: shared hit=972
-> Bitmap Index Scan on idx_users_name_trgm_gist (cost=0.00..32.06 rows=505 width=0) (actual time=19.175..19.175 rows=50000 loops=1)
Index Cond: ((users.name)::text ~~ '%pf%'::text)
Buffers: shared hit=653
Planning time: 0.188 ms
Execution time: 53.231 ms
(11 rows)
以上為個(gè)人經(jīng)驗(yàn)，希望能給大家一個(gè)參考，也希望大家多多支持腳本之家。如有錯(cuò)誤或未考慮完全的地方，望不吝賜教。
您可能感興趣的文章:- PostgreSQL之INDEX 索引詳解
- PostgreSql 重建索引的操作
- postgresql查看表和索引的情況,判斷是否膨脹的操作
- PostgreSQL的B-tree索引用法詳解
- postgresql通過索引優(yōu)化查詢速度操作
- postgresql 索引之 hash的使用詳解
