分布式数据库架构-MyCAT
企业分布式解决方案介绍
360 Atlas-Sharding
Alibaba cobar
Mycat ----> 爱可生 DBLE
TDDL ----> DRDS
Heisenberg ---> 百度
Oceanus ---> 58
Vitess --->
OneProxy
阿里云 DRDS
腾讯 TDSQL分布式策略介绍
Mycat安装及配置文件结构
3.1 预先安装Java运行环境
yum install -y java
3.2下载
Mycat-server-xxxxx.linux.tar.gz
http://dl.mycat.io/
3.3 解压文件
tar xf Mycat-server-1.6.5-release-20180122220033-linux.tar.gz
3.4 软件目录结构
ls
bin catlet conf lib logs version.txt
3.5 启动和连接
配置环境变量
vim /etc/profile
export PATH=/application/mycat/bin:$PATH
source /etc/profile
启动
mycat start
连接mycat：
mysql -uroot -p123456 -h 127.0.0.1 -P8066
3.6 配置文件介绍
logs目录:
wrapper.log ---->mycat启动日志
mycat.log ---->mycat详细工作日志
conf目录:
schema.xml
主配置文件（读写分离、高可用、分布式策略定制、节点控制）
server.xml
mycat软件本身相关的配置
rule.xml
分片规则配置文件,记录分片规则列表、使用方法等
- Mycat的基础功能
4.0 schema.xml 配置文件介绍
db01:
mysql -S /data/3307/mysql.sock
grant all on . to root@'10.0.0.%' identified by '123';
source /root/world.sql
mysql -S /data/3308/mysql.sock
grant all on . to root@'10.0.0.%' identified by '123';
source /root/world.sql
[root@db01 /usr/local/mycat/conf]# mv schema.xml schema.xml.bak
[root@db01 /usr/local/mycat/conf]# vim schema.xml
===================文件头========================
<?xml version="1.0"?>
<!DOCTYPE mycat:schema SYSTEM "schema.dtd">
<mycat:schema xmlns:mycat="http://io.mycat/">
====================================
<schema name="TESTDB" checkSQLschema="false" sqlMaxLimit="100" dataNode="dn1">
</schema>
<dataNode name="dn1" dataHost="localhost1" database= "world" />
<dataHost name="localhost1" maxCon="1000" minCon="10" balance="1" writeType="0" dbType="mysql" dbDriver="native" switchType="1">
<heartbeat>select user()</heartbeat>
<writeHost host="db1" url="10.0.0.51:3307" user="root" password="123">
<readHost host="db2" url="10.0.0.51:3309" user="root" password="123" />
</writeHost>
</dataHost>
</mycat:schema>
4.1 Mycat 读写分离功能
<?xml version="1.0"?>
<!DOCTYPE mycat:schema SYSTEM "schema.dtd">
<mycat:schema xmlns:mycat="http://io.mycat/">
<schema name="TESTDB" checkSQLschema="false" sqlMaxLimit="100" dataNode="dn1">
</schema>
<dataNode name="dn1" dataHost="localhost1" database= "world" />
<dataHost name="localhost1" maxCon="1000" minCon="10" balance="1" writeType="0" dbType="mysql" dbDriver="native" switchType="1">
<heartbeat>select user()</heartbeat>
<writeHost host="db1" url="10.0.0.51:3307" user="root" password="123">
<readHost host="db2" url="10.0.0.51:3309" user="root" password="123" />
</writeHost>
</dataHost>
</mycat:schema>
[root@db01 /usr/local/mycat/conf]# mycat restart
测试:
mysql> select @@server_id;
+-------------+
| @@server_id |
+-------------+
| 9 |
+-------------+
1 row in set (0.00 sec)
mysql> begin; select @@server_id;commit;
Query OK, 0 rows affected (0.00 sec)
[root@db01 /usr/local/mycat/conf]# mysql -uroot -p123456 -h 127.0.0.1 -P8066
4.2 Mycat 读写分离+高可用
[root@db01 conf]# mv schema.xml schema.xml.rw
[root@db01 conf]# vim schema.xml
<?xml version="1.0"?>
<!DOCTYPE mycat:schema SYSTEM "schema.dtd">
<mycat:schema xmlns:mycat="http://io.mycat/">
<schema name="TESTDB" checkSQLschema="false" sqlMaxLimit="100" dataNode="sh1">
</schema>
<dataNode name="sh1" dataHost="oldguo1" database= "world" />
<dataHost name="oldguo1" maxCon="1000" minCon="10" balance="1" writeType="0" dbType="mysql" dbDriver="native" switchType="1">
<heartbeat>select user()</heartbeat>
<writeHost host="db1" url="10.0.0.51:3307" user="root" password="123">
<readHost host="db2" url="10.0.0.51:3309" user="root" password="123" />
</writeHost>
<writeHost host="db3" url="10.0.0.52:3307" user="root" password="123">
<readHost host="db4" url="10.0.0.52:3309" user="root" password="123" />
</writeHost>
</dataHost>
</mycat:schema>
测试:
读写分离:
mysql> select @@server_id;
mysql> begin; select @@server_id;commit;
4.3 自带配置信息介绍
balance属性:
负载均衡类型，目前的取值有3种：
- balance="0", 不开启读写分离机制，所有读操作都发送到当前可用的writeHost上。
- balance="1"，全部的readHost与standby writeHost参与select语句的负载均衡，简单的说，
- balance="2"，所有读操作都随机的在writeHost、readhost上分发。
writeType属性
负载均衡类型，目前的取值有2种：
- writeType="0", 所有写操作发送到配置的第一个writeHost
- writeType="1"，所有写操作都随机的发送到配置的writeHost，但不推荐使用
switchType属性
-1 表示不自动切换
1 默认值，自动切换
2 基于MySQL主从同步的状态决定是否切换 ，心跳语句为 show slave status
datahost其他配置
<dataHost name="localhost1" maxCon="1000" minCon="10" balance="1" writeType="0" dbType="mysql" dbDriver="native" switchType="1">
maxCon="1000"：最大的并发连接数
minCon="10" ：mycat在启动之后，会在后端节点上自动开启的连接线程
tempReadHostAvailable="1"
这个一主一从时（1个writehost，1个readhost时），可以开启这个参数，如果2个writehost，2个readhost时
<heartbeat>select user()</heartbeat> 监测心跳
- Mycat的核心分片技术
5.1 Mycat 垂直分表
mv schema.xml schema.xml.ha
vim schema.xml
<?xml version="1.0"?>
<!DOCTYPE mycat:schema SYSTEM "schema.dtd">
<mycat:schema xmlns:mycat="http://io.mycat/">
<schema name="oldguo" checkSQLschema="false" sqlMaxLimit="100" dataNode="sh1">
<table name="user" dataNode="sh1"/>
<table name="order_t" dataNode="sh2"/>
</schema>
<dataNode name="sh1" dataHost="oldguo1" database= "taobao" />
<dataNode name="sh2" dataHost="oldguo2" database= "taobao" />
<dataHost name="oldguo1" maxCon="1000" minCon="10" balance="1" writeType="0" dbType="mysql" dbDriver="native" switchType="1">
<heartbeat>select user()</heartbeat>
<writeHost host="db1" url="10.0.0.51:3307" user="root" password="123">
<readHost host="db2" url="10.0.0.51:3309" user="root" password="123" />
</writeHost>
<writeHost host="db3" url="10.0.0.52:3307" user="root" password="123">
<readHost host="db4" url="10.0.0.52:3309" user="root" password="123" />
</writeHost>
</dataHost>
<dataHost name="oldguo2" maxCon="1000" minCon="10" balance="1" writeType="0" dbType="mysql" dbDriver="native" switchType="1">
<heartbeat>select user()</heartbeat>
<writeHost host="db1" url="10.0.0.51:3308" user="root" password="123">
<readHost host="db2" url="10.0.0.51:3310" user="root" password="123" />
</writeHost>
<writeHost host="db3" url="10.0.0.52:3308" user="root" password="123">
<readHost host="db4" url="10.0.0.52:3310" user="root" password="123" />
</writeHost>
</dataHost>
</mycat:schema>
创建测试库和表:
mysql -S /data/3307/mysql.sock -e "create database taobao charset utf8;"
mysql -S /data/3308/mysql.sock -e "create database taobao charset utf8;"
mysql -S /data/3307/mysql.sock -e "use taobao;create table user(id int,name varchar(20))";
mysql -S /data/3308/mysql.sock -e "use taobao;create table order_t(id int,name varchar(20))"
测试:
后端节点查看:
[root@db01 /usr/local/mycat/conf]# mysql -S /data/3307/mysql.sock -e "show tables from taobao;"
+------------------+
| Tables_in_taobao |
+------------------+
| user |
+------------------+
[root@db01 /usr/local/mycat/conf]# mysql -S /data/3308/mysql.sock -e "show tables from taobao;"
+------------------+
| Tables_in_taobao |
+------------------+
| order_t |
+------------------+
[root@db01 /usr/local/mycat/conf]#
登录mycat查看现象:
[root@db01 /usr/local/mycat/conf]# mysql -uroot -p123456 -h10.0.0.51 -P8066
mysql> show tables;
+------------------+
| Tables_in_taobao |
+------------------+
| order_t |
| user |
+------------------+
2 rows in set (0.00 sec)
5.2 水平分表
比如说t3表
(1)行数非常多，2000w（1-1000w:sh1 1000w01-2000w:sh2）
(2)访问非常频繁，用户访问较离散
5.2.1 range分片策略应用
vim schema.xml
<?xml version="1.0"?>
<!DOCTYPE mycat:schema SYSTEM "schema.dtd">
<mycat:schema xmlns:mycat="http://io.mycat/">
<schema name="oldguo" checkSQLschema="false" sqlMaxLimit="100" dataNode="sh1">
<table name="user" dataNode="sh1"/>
<table name="order_t" dataNode="sh2"/>
<table name="t3" dataNode="sh1,sh2" rule="auto-sharding-long" />
</schema>
<dataNode name="sh1" dataHost="oldguo1" database= "taobao" />
<dataNode name="sh2" dataHost="oldguo2" database= "taobao" />
<dataHost name="oldguo1" maxCon="1000" minCon="10" balance="1" writeType="0" dbType="mysql" dbDriver="native" switchType="1">
<heartbeat>select user()</heartbeat>
<writeHost host="db1" url="10.0.0.51:3307" user="root" password="123">
<readHost host="db2" url="10.0.0.51:3309" user="root" password="123" />
</writeHost>
<writeHost host="db3" url="10.0.0.52:3307" user="root" password="123">
<readHost host="db4" url="10.0.0.52:3309" user="root" password="123" />
</writeHost>
</dataHost>
<dataHost name="oldguo2" maxCon="1000" minCon="10" balance="1" writeType="0" dbType="mysql" dbDriver="native" switchType="1">
<heartbeat>select user()</heartbeat>
<writeHost host="db1" url="10.0.0.51:3308" user="root" password="123">
<readHost host="db2" url="10.0.0.51:3310" user="root" password="123" />
</writeHost>
<writeHost host="db3" url="10.0.0.52:3308" user="root" password="123">
<readHost host="db4" url="10.0.0.52:3310" user="root" password="123" />
</writeHost>
</dataHost>
</mycat:schema>
查看:
vim rule.xml
<tableRule name="auto-sharding-long">
<rule>
<columns>id</columns>
<algorithm>rang-long</algorithm>
</rule>
<function name="rang-long"
class="io.mycat.route.function.AutoPartitionByLong">
<property name="mapFile">autopartition-long.txt</property>
</function>
===================================
vim autopartition-long.txt
0-10=0
10-20=1
创建测试表：
mysql -S /data/3307/mysql.sock -e "use taobao;create table t3 (id int not null primary key auto_increment,name varchar(20) not null);"
mysql -S /data/3308/mysql.sock -e "use taobao;create table t3 (id int not null primary key auto_increment,name varchar(20) not null);"
测试：
重启mycat
mycat restart
mysql -uroot -p123456 -h 127.0.0.1 -P 8066
use TSETDB
insert into t3(id,name) values(1,'a');
insert into t3(id,name) values(2,'b');
insert into t3(id,name) values(3,'c');
insert into t3(id,name) values(4,'d');
insert into t3(id,name) values(11,'aa');
insert into t3(id,name) values(12,'bb');
insert into t3(id,name) values(13,'cc');
insert into t3(id,name) values(14,'dd');
5.2.2 取模分片mod-long
取余分片方式：分片键（一个列）与节点数量进行取余，得到余数，将数据写入对应节点
vim schema.xml
添加:
<table name="t4" dataNode="sh1,sh2" rule="mod-long" />
vim rule.xml
<property name="count">2</property>
创建测试表：
mysql -S /data/3307/mysql.sock -e "use taobao;create table t4 (id int not null primary key auto_increment,name varchar(20) not null);"
mysql -S /data/3308/mysql.sock -e "use taobao;create table t4 (id int not null primary key auto_increment,name varchar(20) not null);"
重启mycat
mycat restart
测试：
mysql -uroot -p123456 -h10.0.0.51 -P8066
use TESTDB
insert into t4(id,name) values(1,'a');
insert into t4(id,name) values(2,'b');
insert into t4(id,name) values(3,'c');
insert into t4(id,name) values(4,'d');
分别登录后端节点查询数据
mysql -S /data/3307/mysql.sock -e "select * from taobao.t4;"
mysql -S /data/3308/mysql.sock -e "select * from taobao.t4;"
5.2.3 枚举分片
t5 表
id name telnum
1 bj 1212
2 sh 22222
3 bj 3333
4 sh 44444
5 bj 5555
sharding-by-intfile
vim schema.xml
<table name="t5" dataNode="sh1,sh2" rule="sharding-by-intfile" />
vim rule.xml
<tableRule name="sharding-by-intfile">
<rule> <columns>name</columns>
<algorithm>hash-int</algorithm>
</rule>
</tableRule>
<function name="hash-int" class="org.opencloudb.route.function.PartitionByFileMap">
<property name="mapFile">partition-hash-int.txt</property>
<property name="type">1</property>
<property name="defaultNode">0</property>
</function>
partition-hash-int.txt 配置：
bj=0
sh=1
DEFAULT_NODE=1
columns 标识将要分片的表字段，algorithm 分片函数， 其中分片函数配置中，mapFile标识配置文件名称
准备测试环境
mysql -S /data/3307/mysql.sock -e "use taobao;create table t5 (id int not null primary key auto_increment,name varchar(20) not null);"
mysql -S /data/3308/mysql.sock -e "use taobao;create table t5 (id int not null primary key auto_increment,name varchar(20) not null);"
重启mycat
mycat restart
mysql -uroot -p123456 -h10.0.0.51 -P8066
use TESTDB
insert into t5(id,name) values(1,'bj');
insert into t5(id,name) values(2,'sh');
insert into t5(id,name) values(3,'bj');
insert into t5(id,name) values(4,'sh');
insert into t5(id,name) values(5,'tj');
分别登录后端节点查询数据
mysql -S /data/3307/mysql.sock -e "select * from taobao.t5;"
mysql -S /data/3308/mysql.sock -e "select * from taobao.t5;"
5.2.4 Mycat全局表
a b c d .....
join
t
select t1.name ,t.x from t1
join t
select t2.name ,t.x from t2
join t
select t3.name ,t.x from t3
join t
使用场景：
如果你的业务中有些数据类似于数据字典，比如配置文件的配置，
常用业务的配置或者数据量不大很少变动的表，这些表往往不是特别大，
而且大部分的业务场景都会用到，那么这种表适合于Mycat全局表，无须对数据进行切分，
要在所有的分片上保存一份数据即可，Mycat 在Join操作中，业务表与全局表进行Join聚合会优先选择相同分片内的全局表join，
避免跨库Join，在进行数据插入操作时，mycat将把数据分发到全局表对应的所有分片执行，在进行数据读取时候将会随机获取一个节点读取数据。
vim schema.xml
<table name="t_area" primaryKey="id" type="global" dataNode="sh1,sh2" />
后端数据准备
mysql -S /data/3307/mysql.sock
use taobao
create table t_area (id int not null primary key auto_increment,name varchar(20) not null);
mysql -S /data/3308/mysql.sock
use taobao
create table t_area (id int not null primary key auto_increment,name varchar(20) not null);
重启mycat
mycat restart
测试：
mysql -uroot -p123456 -h10.0.0.52 -P8066
use TESTDB
insert into t_area(id,name) values(1,'a');
insert into t_area(id,name) values(2,'b');
insert into t_area(id,name) values(3,'c');
insert into t_area(id,name) values(4,'d');
5.2.5 E-R分片介绍
A
join
B
为了防止跨分片join，可以使用E-R模式
A join B
on a.xx=b.yy
join C
on A.id=C.id
<table name="A" dataNode="sh1,sh2" rule="mod-long">
<childTable name="B" joinKey="yy" parentKey="xx" />
</table>
