今天有朋友問題,MEMORY 引擎的表查詢速度居然比MYISAM引擎慢!
熟讀手冊後,你就不用有這樣的疑問了。
咱們來小解決下。
示例表結構:
create table t1_memory (
id int unsigned not null auto_increment primary key,
a1 decimal(15,12),
a2 decimal(15,12),
remark varchar(200) not null,
key idx_u1 (a1,a2)
) engine memory;
create table t1_myisam (
id int unsigned not null auto_increment primary key,
a1 decimal(15,12),
a2 decimal(15,12),
remark varchar(200) not null,
key idx_u1 (a1,a2)
) engine myisam;
示例SQL語句:
select * from t1_memory where a1>110 and a1<111 and a2>23 and a2<24;
select * from t1_myisam where a1>110 and a1<111 and a2>23 and a2<24;
語句執行計劃:
explain
select * from t1_memory where a1>110 and a1<111 and a2>23 and a2<24;
query result
id |
select_type |
table |
type |
possible_keys |
key |
key_len |
ref |
rows |
Extra |
1 |
SIMPLE |
t1_memory |
ALL |
idx_u1 |
(NULL) |
(NULL) |
(NULL) |
3000 |
Using where |
explain
select * from t1_myisam where a1>110 and a1<111 and a2>23 and a2<24;
query result
id |
select_type |
table |
type |
possible_keys |
key |
key_len |
ref |
rows |
Extra |
1 |
SIMPLE |
t1_myisam |
range |
idx_u1 |
idx_u1 |
9 |
(NULL) |
1 |
Using where |
根本緣由就是默認MEMORY 引擎採用HASH索引, 因此對於RANGE INDEX 來講,咱們要修改爲BTREE索引。
解決辦法:
變化索引類型
alter table t1_memory drop key idx_u1, add key idx_u1 using btree (a1,a2);
優化後執行計劃:
explain
select * from t1_memory where a1>110 and a1<111 and a2>23 and a2<24;
query result
id |
select_type |
table |
type |
possible_keys |
key |
key_len |
ref |
rows |
Extra |
1 |
SIMPLE |
t1_memory |
range |
idx_u1 |
idx_u1 |
9 |
(NULL) |
2 |
Using where |
看到了吧,咱也用上了索引。哈哈。
本文出自 「上帝,我們不見不散!」 博客,請務必保留此出處http://yueliangdao0608.blog.51cto.com/397025/228925優化