“明明给字段建了索引,可带函数的查询还是走全表扫描,数据量大了直接卡到超时……”
如果你在使用 OceanBase 或 MySQL 时遇到过这种问题,大概率是没用到函数索引!当查询条件包含函数计算(比如LOWER(name)、MD5(customer_id))时,普通索引完全失效,而函数索引能直接基于计算结果建立索引,让查询效率飙升10倍+!
今天这篇干货,从定义、用法、避坑到实战场景,带你彻底掌握 OceanBase 函数索引的核心玩法~
函数索引是 基于表达式或函数计算结果创建的索引,而非直接基于列值。
在MySQL 8.0.13开始支持函数索引,主要通过索引计算后的值来提高查询性能。
OceanBase 支持版本:V4.2.0 及以上;
参考文档: MySQL 8.0 参考手册
ps:普通索引为
[索引名](列名),函数索引为
[索引名]((列名));两层括号后开始填写函数表达式;
创建函数索引时,会向主表添加一个隐藏的虚拟生成列,定义为函数索引的索引键,然后再将虚拟生成列的值存储到索引表中。
/* 建表时创建*/
CREATE
TABLE tst_tb (col1
int, col2
int, [
UNIQUE] {
KEY /
INDEX} [索引名]((用户函数(col1)));
/* 使用 create index 创建 */
CREATE [
UNIQUE]
INDEX tst_tb_func_idx2
ON tst_tb((col1+
3));
/* 使用 alter table 创建 */
ALTER
TABLE tst_tb
ADD [
UNIQUE] {
INDEX |
KEY} [索引名] ((用户函数(col1)));
/*
示例1:对名字的小写形式创建索引,使用与索引定义完全相同的表达式查询
*/
CREATE
INDEX idx_lower_name
ON tst_tb ((
LOWER(
name)));
SELECT *
FROM employees
WHERE
LOWER(
name) =
'John';
/*
示例2:对JSON字段的特定路径创建索引
从 JSON 文档中指定的路径返回数据
*/
CREATE
INDEX idx_json_extract
ON tst_tb ((JSON_EXTRACT(person,
'$.weight')));
SELECT *
FROM tst_tb
WHERE JSON_EXTRACT(specs,
'$.weight') > 10;
> 1. 非确定性的系统函数禁止被用于创建函数索引;
> 2. 子查询、参数、变量、存储过程、自定义函数 function 不允许在函数索引上使用;
> 3. 函数索引的表达式结果受限于字段长度;
> 4. 函数索引无法作为主键使用;
> 5. 空间索引和全文索引不能定义为函数索引;
> 6. 函数索引不允许在外键中使用;
> 7. 某字段用于函数索引,删列前必须删除该索引,否则报错;
/* 1. 使用非确定性函数*/
create
index tst_01
on tst_tb((
now()));
失败原因:ErrorCode = 3758, SQLState = HY000, Details = Expression of functional index contains a disallowed function.
/*在创建函数索引时,使用了不支持的函数,索引函数使用 SYSDATE 或用户环境*/
create
index current_time_fun_idx
on tst_tb((
current_date()));
失败原因:ErrorCode = 3758, SQLState = HY000, Details = Expression of functional index contains a disallowed function.
/* 2.表达式过于复杂,函数索引中包括了子查询*/
CREATE
INDEX tst_02
ON tst_tb ((
SELECT
MAX(
id)
FROM tst_tb));
失败原因:ErrorCode = 1064, SQLState = 42000, Details = You have an error in your SQL syntax;
check the
manual that corresponds
to your OceanBase
version
for the
right syntax
to
use near
'SELECT MAX(id) FROM tst_tb))'
at line
1
/* 3.表达式结果太长,REPEAT 重复多次改字段字符串拼接*/
CREATE
INDEX idx_invalid
ON tst_tb ((
REPEAT(customer_id,
100)));
失败原因:ErrorCode = 3757, SQLState = HY000, Details = Cannot
create a functional
index
on an expression that
returns a
BLOB
or TEXT. Please
consider
using CAST.
/* 在创建函数索引时,使用了返回 BLOB 或 TEXT 类型的表达式,而 BLOB 或 TEXT 类型的表达式不能用于创建函数索引 */
/* 4.函数索引无法作为主键创建*/
CREATE
TABLE
`tst_tb` (
`id`
bigint(
20)
unsigned
NOT
NULL AUTO_INCREMENT
COMMENT
'主键id,自动增长',
`customer_id`
varchar(
50)
DEFAULT
NULL
COMMENT
'客户ID',
`create_time` datetime
NOT
NULL
DEFAULT
CURRENT_TIMESTAMP
COMMENT
'创建时间',
`update_time` datetime
DEFAULT
NULL
ON
UPDATE
CURRENT_TIMESTAMP
COMMENT
'更新时间',
`product`
varchar(
10)
DEFAULT
NULL
COMMENT
'产品类型',
PRIMARY
KEY
`pri_cust` ((
lower(
md5(
concat(
upper(customer_id),
'oceanbase'))))),
KEY
`idx_create_time` (
`create_time`) BLOCK_SIZE
16384
LOCAL,
KEY
`ob_idx_customer_id` (
`customer_id`) BLOCK_SIZE
16384
LOCAL
);
失败原因:ErrorCode = 1064, SQLState = 42000, Details = You have an error in your SQL syntax;
check the
manual that corresponds
to your OceanBase
version
for the
right syntax
to
use near
'(lower(md5(concat(upper(customer_id), 'oceanbase
'))))), KEY `idx_create_time` '
at line
7
/* 5.空间索引和全文索引不能定义为函数索引*/
CREATE
TABLE tbl_func (c1
int, c2
int,
index tbl1_func_idx1 ((c1+
1)), SPATIAL
KEY ((c1+c2)));
失败原因:ErrorCode = 3760, SQLState = HY000, Details = Spatial functional index is not supported.
/* 不支持空间索引套用函数索引*/
CREATE
TABLE tbl_func (c1
int, c2
varchar(
20),
index tbl1_func_idx1 ((c1+
1)), FULLTEXT
INDEX full_idx_func ((
SUBSTR(c2,
1,
2)))
WITH PARSER ngram PARSER_PROPERTIES=(ngram_token_size=
3));
失败原因:ErrorCode = 1235, SQLState = 0A000, Details = Fulltext index on generated column is not supported
/* 不支持全文索引套用函数索引*/
/* 6.函数索引不允许在外键中使用*/
CREATE
TABLE products (
product_id
INT PRIMARY
KEY,
product_code
VARCHAR(
20)
NOT
NULL,
product_name
VARCHAR(
100)
NOT
NULL,
INDEX idx_lower_product_code ((
LOWER(product_code)))
);
CREATE
TABLE orders (
order_id
INT PRIMARY
KEY,
product_code_ref
VARCHAR(
20),
CONSTRAINT fk_order_product
FOREIGN
KEY (product_code_ref)
-- 创建外键引用函数索引
REFERENCES products (
LOWER(product_code))
);
失败原因:ErrorCode = 1064, SQLState = 42000, Details = You have an error in your SQL syntax;
check the
manual that corresponds
to your OceanBase
version
for the
right syntax
to
use near
'(product_code)) )'
at line
7
/* 7. 某字段用于函数索引,删列前必须删除该索引,否则报错*/
ALTER
TABLE products
DROP
COLUMN product_code;
失败原因:ErrorCode = 3837, SQLState = HY000, Details = Column 'product_code' has a functional index dependency and cannot be dropped or renamed.
参考文档:
/*表定义语句*/
CREATE
TABLE
`tst_tb` (
`id`
bigint(
20)
unsigned
NOT
NULL AUTO_INCREMENT
COMMENT
'主键id,自动增长',
`customer_id`
varchar(
50)
DEFAULT
NULL
COMMENT
'客户ID',
`create_time` datetime
NOT
NULL
DEFAULT
CURRENT_TIMESTAMP
COMMENT
'创建时间',
`update_time` datetime
DEFAULT
NULL
ON
UPDATE
CURRENT_TIMESTAMP
COMMENT
'更新时间',
`product`
varchar(
10)
DEFAULT
NULL
COMMENT
'产品类型',
`name`
varchar(
100)
DEFAULT
NULL,
PRIMARY
KEY (
`id`),
KEY
`idx_create_time` (
`create_time`) BLOCK_SIZE
16384
LOCAL,
KEY
`ob_idx_customer_id` (
`customer_id`) BLOCK_SIZE
16384
LOCAL
);
explain
extended
select *
from tst_tb
where
name=
'Product One';
=================================================
|ID|OPERATOR |NAME |EST.ROWS|EST.TIME(us)|
-------------------------------------------------
|0 |TABLE FULL SCAN|tst_tb|1 |3 |
=================================================
Outputs & filters:
-------------------------------------
0 - output([tst_tb.id(0x7ff8f1823480)], [tst_tb.customer_id(0x7ff8f18238d0)], [tst_tb.create_time(0x7ff8f1823d20)], [tst_tb.update_time(0x7ff8f1824170)],
[tst_tb.product(0x7ff8f18245c0)], [tst_tb.name(0x7ff8f1822d60)]), filter([tst_tb.name(0x7ff8f1822d60) = 'Product One'(0x7ff8f1822000)(0x7ff8f1822490)]), rowset=16
access([tst_tb.id(0x7ff8f1823480)], [tst_tb.name(0x7ff8f1822d60)], [tst_tb.customer_id(0x7ff8f18238d0)], [tst_tb.create_time(0x7ff8f1823d20)], [tst_tb.update_time(0x7ff8f1824170)],
[tst_tb.product(0x7ff8f18245c0)]), partitions(p0)
is_index_back=false, is_global_index=false, filter_before_indexback[false],
range_key([tst_tb.id(0x7ff8f1823480)]), range(MIN ; MAX)always true
/* 添加虚拟生成列 */
ALTER
TABLE tst_tb
ADD
COLUMN lower_name
VARCHAR(
100)
GENERATED
ALWAYS
AS (
LOWER(
name))
STORED;
/* 在虚拟生成列上创建索引 */
CREATE
INDEX idx_lower_name
ON tst_tb (lower_name);
/* 执行计划 */
explain
extended
select *
from tst_tb
where lower_name=
'Product One';
==================================================================
|ID|OPERATOR |NAME |EST.ROWS|EST.TIME(us)|
------------------------------------------------------------------
|0 |TABLE RANGE SCAN|tst_tb(idx_lower_name)|1 |8 |
==================================================================
Outputs & filters:
-------------------------------------
0 - output([tst_tb.id(0x7ff8bd22d120)], [tst_tb.customer_id(0x7ff8bd22d570)], [tst_tb.create_time(0x7ff8bd22d9c0)], [tst_tb.update_time(0x7ff8bd22de10)],
[tst_tb.product(0x7ff8bd22e260)], [tst_tb.name(0x7ff8bd223840)], [tst_tb.lower_name(0x7ff8bd221bc0)]), filter(nil), rowset=16
access([tst_tb.id(0x7ff8bd22d120)], [tst_tb.name(0x7ff8bd223840)], [tst_tb.lower_name(0x7ff8bd221bc0)], [tst_tb.customer_id(0x7ff8bd22d570)], [tst_tb.create_time(0x7ff8bd22d9c0)],
[tst_tb.update_time(0x7ff8bd22de10)], [tst_tb.product(0x7ff8bd22e260)]), partitions(p0)
is_index_back=true, is_global_index=false,
range_key([tst_tb.lower_name(0x7ff8bd221bc0)], [tst_tb.id(0x7ff8bd22d120)]), range(Product One,MIN ; Product One,MAX),
range_cond([tst_tb.lower_name(0x7ff8bd221bc0) = 'Product One'(0x7ff8bd2eb070)(0x7ff8bd2ea650)])
表数据

未走索引

走索引

场景描述:客户有两张表,A 表是在线热更数据的,B 表是离线异步更新的,其中 A表中的
customer_id字段是应用程序加密过的字段值,B表中的
customer_id字段是原始数据,客户将该两张表创建了联合查询的视图,为了保持一致,对 B 表的
customer_id 字段做了 md5 加密,但查询视图的时候 B 表没有走索引;
集群版本: OceanBase V4.2.1.10
/*
A 表结构
B 表与A表结构一致,且B表的数据是异步更新
*/
CREATE
TABLE
`tst_ta` (
`id`
bigint(
20)
unsigned
NOT
NULL AUTO_INCREMENT
COMMENT
'主键id,自动增长',
`customer_id`
varchar(
50)
DEFAULT
NULL
COMMENT
'客户ID',
`create_time` datetime
NOT
NULL
DEFAULT
CURRENT_TIMESTAMP
COMMENT
'创建时间',
`update_time` datetime
DEFAULT
NULL
ON
UPDATE
CURRENT_TIMESTAMP
COMMENT
'更新时间',
`product`
varchar(
10)
DEFAULT
NULL
COMMENT
'产品类型',
PRIMARY
KEY (
`id`,
`create_time`),
KEY
`idx_create_time` (
`create_time`) BLOCK_SIZE
16384
LOCAL,
KEY
`ob_idx_customer_id` (
`customer_id`) BLOCK_SIZE
16384
LOCAL
);
测试表模拟数据
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST001',
'2023-01-15 10:30:00',
'A');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST002',
'2023-01-16 11:45:00',
'B');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST003',
'2023-01-17 09:15:00',
'C');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST004',
'2023-01-18 14:20:00',
'A');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST005',
'2023-01-19 16:30:00',
'B');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST006',
'2023-01-20 10:10:00',
'D');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST007',
'2023-01-21 13:25:00',
'A');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST008',
'2023-01-22 15:40:00',
'C');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST009',
'2023-01-23 08:50:00',
'B');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST010',
'2023-01-24 17:00:00',
'D');
INSERT
INTO
`tst_tb` (
`customer_id`,
`product`)
VALUES (
'CUST011',
'E');
INSERT
INTO
`tst_tb` (
`customer_id`,
`product`)
VALUES (
'CUST012',
'A');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST013',
'2023-01-27 12:15:00',
'B');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST014',
'2023-01-28 14:30:00',
'C');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST015',
'2023-01-29 09:45:00',
'D');
INSERT
INTO
`tst_tb` (
`customer_id`,
`product`)
VALUES (
'CUST016',
'A');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST017',
'2023-01-31 11:20:00',
'B');
INSERT
INTO
`tst_tb` (
`customer_id`,
`product`)
VALUES (
'CUST018',
'E');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST019',
'2023-02-02 16:10:00',
'C');
INSERT
INTO
`tst_tb` (
`customer_id`,
`update_time`,
`product`)
VALUES (
'CUST020',
'2023-02-03 10:25:00',
'D');
tst_ta.customer_id 加密后的表数据

tst_tb.customer_id 加密前的表数据

/*
将 customer_id 字段的值统一大写后与'oceanbase'拼接,然后经过 md5 算法加密后统一小写后记录
*/
CREATE
VIEW
`tst_view`
AS
(
select
id,customer_id,create_time,update_time,product
from tst_ta)
union
all
(
select
id,
lower(
md5(
concat(
upper(customer_id),
'oceanbase')))
AS customer_id,
create_time,update_time,product
from tst_tb
);
未走索引

/*
CREATE TABLE `tst_ta` (
`id` bigint(20) unsigned NOT NULL AUTO_INCREMENT COMMENT '主键id,自动增长',
`customer_id` varchar(50) DEFAULT NULL COMMENT '客户ID',
`create_time` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '创建时间',
`update_time` datetime DEFAULT NULL ON UPDATE CURRENT_TIMESTAMP COMMENT '更新时间',
`product` varchar(10) DEFAULT NULL COMMENT '产品类型',
PRIMARY KEY (`id`, `create_time`),
KEY `idx_create_time` (`create_time`) BLOCK_SIZE 16384 LOCAL,
KEY `ob_idx_customer_id` (`customer_id`) BLOCK_SIZE 16384 LOCAL
);
*/
explain
extended
select *
from tst_view
where customer_id=
'366330e31189acd28639de9dd7b29cb1';
========================================================================
|ID|OPERATOR |NAME |EST.ROWS|EST.TIME(us)|
------------------------------------------------------------------------
|0 |UNION ALL | |2 |11 |
|1 |├─TABLE RANGE SCAN|tst_ta(ob_idx_customer_id)|1 |7 |
|2 |└─TABLE FULL SCAN |tst_tb |1 |4 |
========================================================================
Outputs & filters:
-------------------------------------
0 - output([UNION([1])(0x7ff9ce07efa0)], [UNION([2])(0x7ff9ce07f360)], [UNION([3])(0x7ff9ce07f720)], [UNION([4])(0x7ff9ce07fae0)], [UNION([5])(0x7ff9ce07fea0)]), filter(nil), rowset=16
1 - output([tst_ta.id(0x7ff9ce064850)], [tst_ta.customer_id(0x7ff9ce065120)], [tst_ta.create_time(0x7ff9ce0659f0)], [tst_ta.update_time(0x7ff9ce0662c0)],
[tst_ta.product(0x7ff9ce066b90)]), filter(nil), rowset=16
access([tst_ta.__pk_increment(0x7ff9ce084030)], [tst_ta.id(0x7ff9ce064850)], [tst_ta.customer_id(0x7ff9ce065120)], [tst_ta.create_time(0x7ff9ce0659f0)],
[tst_ta.update_time(0x7ff9ce0662c0)], [tst_ta.product(0x7ff9ce066b90)]), partitions(p0)
is_index_back=true, is_global_index=false,
range_key([tst_ta.customer_id(0x7ff9ce065120)], [tst_ta.__pk_increment(0x7ff9ce084030)]), range(366330e31189acd28639de9dd7b29cb1,MIN ; 366330e31189acd28639de9dd7b29cb1,
MAX),
range_cond([tst_ta.customer_id(0x7ff9ce065120) = '366330e31189acd28639de9dd7b29cb1'(0x7ff8f0484a00)(0x7ff8f0483fe0)])
2 - output([tst_tb.id(0x7ff9ce0797e0)], [lower(md5(concat(upper(tst_tb.customer_id(0x7ff9ce07ce60))(0x7ff9ce07b6b0), 'oceanbase')(0x7ff9ce07ade0))(0x7ff9ce07a510))(0x7ff9ce079c40)],
[tst_tb.create_time(0x7ff9ce07d870)], [tst_tb.update_time(0x7ff9ce07e140)], [tst_tb.product(0x7ff9ce07ea10)]), filter([lower(md5(concat(upper(tst_tb.customer_id(0x7ff9ce07ce60))(0x7ff9ce07b6b0),
'oceanbase')(0x7ff9ce07ade0))(0x7ff9ce07a510))(0x7ff9ce079c40) = '366330e31189acd28639de9dd7b29cb1'(0x7ff9ce0813f0)(0x7ff9ce0991d0)]), rowset=16
access([tst_tb.id(0x7ff9ce0797e0)], [tst_tb.create_time(0x7ff9ce07d870)], [tst_tb.customer_id(0x7ff9ce07ce60)], [tst_tb.update_time(0x7ff9ce07e140)],
[tst_tb.product(0x7ff9ce07ea10)]), partitions(p0)
is_index_back=false, is_global_index=false, filter_before_indexback[false],
range_key([tst_tb.id(0x7ff9ce0797e0)], [tst_tb.create_time(0x7ff9ce07d870)]), range(MIN,MIN ; MAX,MAX)always true
走索引

/*添加索引 1*/
create
index tst_tb_fun_idx
on tst_tb((
lower(
md5(
concat(
upper(customer_id),
'oceanbase')))));
/*
执行 select * from tst_view where customer_id='366330e31189acd28639de9dd7b29cb1';
相当于
select id,customer_id,create_time,update_time,product from tst_ta where customer_id='366330e31189acd28639de9dd7b29cb1'
union all
select id,
lower(md5(concat(upper(customer_id), 'oceanbase'))) AS customer_id,
create_time,update_time,product from tst_tb where lower(md5(concat(upper(customer_id), 'oceanbase')))='366330e31189acd28639de9dd7b29cb1';
而不是 select * from tst_tb where customer_id='366330e31189acd28639de9dd7b29cb1';
*/
/* 执行计划 */
explain
extended
select *
from tst_view
where customer_id=
'366330e31189acd28639de9dd7b29cb1';
============================================================================
|ID|OPERATOR |NAME |EST.ROWS|EST.TIME(us)|
----------------------------------------------------------------------------
|0 |EXCHANGE IN REMOTE | |2 |22 |
|1 |└─EXCHANGE OUT REMOTE | |2 |19 |
|2 | └─UNION ALL | |2 |14 |
|3 | ├─TABLE RANGE SCAN|tst_ta(ob_idx_customer_id)|1 |7 |
|4 | └─TABLE RANGE SCAN|tst_tb(tst_tb_fun_idx) |1 |8 |
============================================================================
Outputs & filters:
-------------------------------------
0 - output([UNION([1])(0x7fd63d894110)], [UNION([2])(0x7fd63d8944d0)], [UNION([3])(0x7fd63d894890)], [UNION([4])(0x7fd63d894c50)], [UNION([5])(0x7fd63d895010)]), filter(nil)
1 - output([UNION([1])(0x7fd63d894110)], [UNION([2])(0x7fd63d8944d0)], [UNION([3])(0x7fd63d894890)], [UNION([4])(0x7fd63d894c50)], [UNION([5])(0x7fd63d895010)]), filter(nil)
2 - output([UNION([1])(0x7fd63d894110)], [UNION([2])(0x7fd63d8944d0)], [UNION([3])(0x7fd63d894890)], [UNION([4])(0x7fd63d894c50)], [UNION([5])(0x7fd63d895010)]), filter(nil), rowset=16
3 - output([tst_ta.id(0x7fd63d864850)], [tst_ta.customer_id(0x7fd63d865120)], [tst_ta.create_time(0x7fd63d8659f0)], [tst_ta.update_time(0x7fd63d8662c0)],
[tst_ta.product(0x7fd63d866b90)]), filter(nil), rowset=16
access([tst_ta.__pk_increment(0x7fd63d8a5180)], [tst_ta.id(0x7fd63d864850)], [tst_ta.customer_id(0x7fd63d865120)], [tst_ta.create_time(0x7fd63d8659f0)],
[tst_ta.update_time(0x7fd63d8662c0)], [tst_ta.product(0x7fd63d866b90)]), partitions(p0)
is_index_back=true, is_global_index=false,
range_key([tst_ta.customer_id(0x7fd63d865120)], [tst_ta.__pk_increment(0x7fd63d8a5180)]), range(366330e31189acd28639de9dd7b29cb1,MIN ; 366330e31189acd28639de9dd7b29cb1,
MAX),
range_cond([tst_ta.customer_id(0x7fd63d865120) = '366330e31189acd28639de9dd7b29cb1'(0x7fd76aeda000)(0x7fd76aed95e0)])
4 - output([tst_tb.id(0x7fd63d886840)], [lower(md5(concat(upper(tst_tb.customer_id(0x7fd63d87ddf0))(0x7fd63d888710), 'oceanbase')(0x7fd63d887e40))(0x7fd63d887570))(0x7fd63d886ca0)],
[tst_tb.create_time(0x7fd63d88a480)], [tst_tb.update_time(0x7fd63d88ad50)], [tst_tb.product(0x7fd63d88b620)]), filter(nil), rowset=16
access([tst_tb.id(0x7fd63d886840)], [tst_tb.create_time(0x7fd63d88a480)], [tst_tb.customer_id(0x7fd63d87ddf0)], [tst_tb.update_time(0x7fd63d88ad50)],
[tst_tb.product(0x7fd63d88b620)]), partitions(p0)
is_index_back=true, is_global_index=false,
range_key([tst_tb.SYS_NC21$(0x7fd63d879390)], [tst_tb.id(0x7fd63d886840)], [tst_tb.create_time(0x7fd63d88a480)]), range(366330e31189acd28639de9dd7b29cb1,
MIN,MIN ; 366330e31189acd28639de9dd7b29cb1,MAX,MAX),
range_cond([tst_tb.SYS_NC21$(0x7fd63d879390) = '366330e31189acd28639de9dd7b29cb1'(0x7fd76aedaeb0)(0x7fd76aeda490)])
tst_tb.customer_id='...'的执行计划
tst_tb.lower(md5(concat(upper(customer_id), 'oceanbase')))='...'的执行计划
以上就是 OceanBase 函数索引的核心内容, 如果你在操作中遇到问题,或是有独到的优化建议,欢迎在评论区留言分享,后续我们还会分享更多OceanBase / MySQL 性能优化秘籍,下次见~
更多技术分享可关注
来自 “ ITPUB博客 ” ,链接:https://blog.itpub.net/70047467/viewspace-3105536/ ,如需转载,请注明出处,否则将追究法律责任。