软件界有只猫,不用我说,各位看官肯定知道是哪只,那就是大名鼎鼎的Tomcat,现在又来了一只猫,据说是位东方萌妹子,暂且认作Tom猫的表妹,本来叫OpencloudDB,后又改名为Mycat,或许Cat更亲切?那现在就来认识下这只小猫吧。
数据库的核心地位就不说了,但现在的问题是,各种RDB,各种NoSQL交织,又是分布式、多租户的场景下,心里有没有十足的把握能稳住如此局面呢。有需求,就有市场!自然,相应的技术也应运而生,Mycat作为一款DB中间件,可以作为应用和DB间的“桥梁”,让后台DB的复杂组成对应用透明,处理分库分表、多租户架构和大数据实时查询等都不在话下!
工具:
Idea201902/JDK11/ZK3.5.5/Gradle5.4.1/RabbitMQ3.7.13/Lombok0.26/Erlang21.2/RocketMQ4.5.2/Sentinel1.6.3/SpringBoot2.1.6/RHEL7.6/VMware15.0.4/Mysql8.0.17/Mycat1.6.7.3/MysqlWorkbench6.3
难度: 新手--战士--老兵--大师
目标:
1.Linux下使用Mycat连接Mysql集群(两主一从),读写分离式应用
步骤:
1.建立Mysql集群,步骤参考往期文章(Linux下Mysql集群使用)。
2.下载Mycat,放到Linux中/usr/mycat下,记得先建立此目录。
3.进入该目录,解压:
[[email protected] mycat]# tar -zxvf Mycat-server-1.6.7.3-release-20190828135747-linux.tar.gz
4.可以看到目录结构如下,两猫确实略像:
- bin—Mycat的各种管理命令;
- catlet—扩展功能;
- conf—配置信息,这个也是本期重点使用的;
- lib—jar包库,因为Mycat是Java开发的;
- logs—日志文件;
5.Mycat融合应用的架构,即本次目标架构:
如果需要做扩展高可用,即可变成这样的:
就是这么简单!
6.其实Mycat从应用上讲,就是做配置,源码可按喜好研究,据说很复杂!
主要是三个文件核心文件rule.xml、schema.xml、server.xml的配置:
- server.xml:Mycat的配置文件,可以将Mycat视为DBServer的代理,
- schema.xml:逻辑表与物理DB/分片分库的映射配置,
- rule.xml:分库分表规则,
7.挨个看看长啥样,参数的含义注释上基本有说明,这里都是全局配置参数:
<system>
中设置Mycat全局属性;<firewall>
设置黑白名单;<user>
设置用户登录Mycat的账号信息;<privileges>
单独设置表的DML权限;
server.xml原版样例:
1 <?xml version="1.0" encoding="UTF-8"?> 2 <!DOCTYPE mycat:server SYSTEM "server.dtd"> 3 <mycat:server xmlns:mycat="http://io.mycat/"> 4 <system> 5 <property name="nonePasswordLogin">0</property> <!-- 0为需要密码登陆、1为不需要密码登陆 ,默认为0,设置为1则需要指定默认账户--> 6 <property name="useHandshakeV10">1</property> 7 <property name="useSqlStat">0</property> <!-- 1为开启实时统计、0为关闭 --> 8 <property name="useGlobleTableCheck">0</property> <!-- 1为开启全加班一致性检测、0为关闭 --> 9 <property name="sqlExecuteTimeout">300</property> <!-- SQL 执行超时 单位:秒--> 10 <property name="sequnceHandlerType">2</property> 11 <!--<property name="sequnceHandlerPattern">(?:(\s*next\s+value\s+for\s*MYCATSEQ_(\w+))(,|\)|\s)*)+</property>--> 12 <!--必须带有MYCATSEQ_或者 mycatseq_进入序列匹配流程 注意MYCATSEQ_有空格的情况--> 13 <property name="sequnceHandlerPattern">(?:(\s*next\s+value\s+for\s*MYCATSEQ_(\w+))(,|\)|\s)*)+</property> 14 <property name="subqueryRelationshipCheck">false</property> <!-- 子查询中存在关联查询的情况下,检查关联字段中是否有分片字段 .默认 false --> 15 <!-- <property name="useCompression">1</property>--> <!--1为开启mysql压缩协议--> 16 <!-- <property name="fakeMySQLVersion">5.6.20</property>--> <!--设置模拟的MySQL版本号--> 17 <!-- <property name="processorBufferChunk">40960</property> --> 18 <!-- 19 <property name="processors">1</property> 20 <property name="processorExecutor">32</property> 21 --> 22 <!--默认为type 0: DirectByteBufferPool | type 1 ByteBufferArena | type 2 NettyBufferPool --> 23 <property name="processorBufferPoolType">0</property> 24 <!--默认是65535 64K 用于sql解析时最大文本长度 --> 25 <!--<property name="maxStringLiteralLength">65535</property>--> 26 <!--<property name="sequnceHandlerType">0</property>--> 27 <!--<property name="backSocketNoDelay">1</property>--> 28 <!--<property name="frontSocketNoDelay">1</property>--> 29 <!--<property name="processorExecutor">16</property>--> 30 <!-- 31 <property name="serverPort">8066</property> <property name="managerPort">9066</property> 32 <property name="idleTimeout">300000</property> <property name="bindIp">0.0.0.0</property> 33 <property name="dataNodeIdleCheckPeriod">300000</property> 5 * 60 * 1000L; //连接空闲检查 34 <property name="frontWriteQueueSize">4096</property> <property name="processors">32</property> --> 35 <!--分布式事务开关,0为不过滤分布式事务,1为过滤分布式事务(如果分布式事务内只涉及全局表,则不过滤),2为不过滤分布式事务,但是记录分布式事务日志--> 36 <property name="handleDistributedTransactions">0</property> 37 <!--off heap for merge/order/group/limit 1开启 0关闭--> 38 <property name="useOffHeapForMerge">0</property> 39 <!--单位为m--> 40 <property name="memoryPageSize">64k</property> 41 <!--单位为k--> 42 <property name="spillsFileBufferSize">1k</property> 43 <property name="useStreamOutput">0</property> 44 <!--单位为m--> 45 <property name="systemReserveMemorySize">384m</property> 46 <!--是否采用zookeeper协调切换 --> 47 <property name="useZKSwitch">false</property> 48 <!-- XA Recovery Log日志路径 --> 49 <!--<property name="XARecoveryLogBaseDir">./</property>--> 50 <!-- XA Recovery Log日志名称 --> 51 <!--<property name="XARecoveryLogBaseName">tmlog</property>--> 52 <!--如果为 true的话 严格遵守隔离级别,不会在仅仅只有select语句的时候在事务中切换连接--> 53 <property name="strictTxIsolation">false</property> 54 <property name="useZKSwitch">true</property> 55 </system> 56 57 <!-- 全局SQL防火墙设置 --> 58 <!--白名单可以使用通配符%或着*--> 59 <!--例如<host host="127.0.0.*" user="root"/>--> 60 <!--例如<host host="127.0.*" user="root"/>--> 61 <!--例如<host host="127.*" user="root"/>--> 62 <!--例如<host host="1*7.*" user="root"/>--> 63 <!--这些配置情况下对于127.0.0.1都能以root账户登录--> 64 <!-- 65 <firewall> 66 <whitehost> 67 <host host="1*7.0.0.*" user="root"/> 68 </whitehost> 69 <blacklist check="false"> 70 </blacklist> 71 </firewall> 72 --> 73 <user name="root" defaultAccount="true"> 74 <property name="password">123456</property> 75 <property name="schemas">TESTDB</property> 76 <!-- 表级 DML 权限设置 --> 77 <!-- 78 <privileges check="false"> 79 <schema name="TESTDB" dml="0110" > 80 <table name="tb01" dml="0000"></table> 81 <table name="tb02" dml="1111"></table> 82 </schema> 83 </privileges> 84 --> 85 </user> 86 <user name="user"> 87 <property name="password">user</property> 88 <property name="schemas">TESTDB</property> 89 <property name="readOnly">true</property> 90 </user> 91 </mycat:server>
8.schema.xml,配置schema下各个table的分片/分库,以及物理DB:
<schema>+<table>
租户和子表配置,<dataNode>
分片,<dataHost>
物理DB,
原版样例:
1 <?xml version="1.0"?> 2 <!DOCTYPE mycat:schema SYSTEM "schema.dtd"> 3 <mycat:schema xmlns:mycat="http://io.mycat/"> 4 <schema name="TESTDB" checkSQLschema="true" sqlMaxLimit="100"> 5 <!-- auto sharding by id (long) --> 6 <table name="travelrecord" dataNode="dn1,dn2,dn3" rule="auto-sharding-long" /> 7 <!-- global table is auto cloned to all defined data nodes ,so can join 8 with any table whose sharding node is in the same data node --> 9 <table name="company" primaryKey="ID" type="global" dataNode="dn1,dn2,dn3" /> 10 <table name="goods" primaryKey="ID" type="global" dataNode="dn1,dn2" /> 11 <!-- random sharding using mod sharind rule --> 12 <table name="hotnews" primaryKey="ID" autoIncrement="true" dataNode="dn1,dn2,dn3" 13 rule="mod-long" /> 14 <!-- <table name="dual" primaryKey="ID" dataNode="dnx,dnoracle2" type="global" 15 needAddLimit="false"/> <table name="worker" primaryKey="ID" dataNode="jdbc_dn1,jdbc_dn2,jdbc_dn3" 16 rule="mod-long" /> --> 17 <table name="employee" primaryKey="ID" dataNode="dn1,dn2" 18 rule="sharding-by-intfile" /> 19 <table name="customer" primaryKey="ID" dataNode="dn1,dn2" 20 rule="sharding-by-intfile"> 21 <childTable name="orders" primaryKey="ID" joinKey="customer_id" 22 parentKey="id"> 23 <childTable name="order_items" joinKey="order_id" 24 parentKey="id" /> 25 </childTable> 26 <childTable name="customer_addr" primaryKey="ID" joinKey="customer_id" 27 parentKey="id" /> 28 </table> 29 <!-- <table name="oc_call" primaryKey="ID" dataNode="dn1$0-743" rule="latest-month-calldate" 30 /> --> 31 </schema> 32 <!-- <dataNode name="dn1$0-743" dataHost="localhost1" database="db$0-743" 33 /> --> 34 <dataNode name="dn1" dataHost="localhost1" database="db1" /> 35 <dataNode name="dn2" dataHost="localhost1" database="db2" /> 36 <dataNode name="dn3" dataHost="localhost1" database="db3" /> 37 <!--<dataNode name="dn4" dataHost="sequoiadb1" database="SAMPLE" /> 38 <dataNode name="jdbc_dn1" dataHost="jdbchost" database="db1" /> 39 <dataNode name="jdbc_dn2" dataHost="jdbchost" database="db2" /> 40 <dataNode name="jdbc_dn3" dataHost="jdbchost" database="db3" /> --> 41 <dataHost name="localhost1" maxCon="1000" minCon="10" balance="0" 42 writeType="0" dbType="mysql" dbDriver="native" switchType="1" slaveThreshold="100"> 43 <heartbeat>select user()</heartbeat> 44 <!-- can have multi write hosts --> 45 <writeHost host="hostM1" url="localhost:3306" user="root" 46 password="123456"> 47 <!-- can have multi read hosts --> 48 <readHost host="hostS2" url="192.168.1.200:3306" user="root" password="xxx" /> 49 </writeHost> 50 <writeHost host="hostS1" url="localhost:3316" user="root" 51 password="123456" /> 52 <!-- <writeHost host="hostM2" url="localhost:3316" user="root" password="123456"/> --> 53 </dataHost> 54 <!-- 55 <dataHost name="sequoiadb1" maxCon="1000" minCon="1" balance="0" dbType="sequoiadb" dbDriver="jdbc"> 56 <heartbeat> </heartbeat> 57 <writeHost host="hostM1" url="sequoiadb://1426587161.dbaas.sequoialab.net:11920/SAMPLE" user="jifeng" password="jifeng"></writeHost> 58 </dataHost> 59 ? 60 <dataHost name="oracle1" maxCon="1000" minCon="1" balance="0" writeType="0" dbType="oracle" dbDriver="jdbc"> <heartbeat>select 1 from dual</heartbeat> 61 <connectionInitSql>alter session set nls_date_format=‘yyyy-mm-dd hh24:mi:ss‘</connectionInitSql> 62 <writeHost host="hostM1" url="jdbc:oracle:thin:@127.0.0.1:1521:nange" user="base" password="123456" > </writeHost> </dataHost> 63 ? 64 <dataHost name="jdbchost" maxCon="1000" minCon="1" balance="0" writeType="0" dbType="mongodb" dbDriver="jdbc"> 65 <heartbeat>select user()</heartbeat> 66 <writeHost host="hostM" url="mongodb://192.168.0.99/test" user="admin" password="123456" ></writeHost> </dataHost> 67 ? 68 <dataHost name="sparksql" maxCon="1000" minCon="1" balance="0" dbType="spark" dbDriver="jdbc"> 69 <heartbeat> </heartbeat> 70 <writeHost host="hostM1" url="jdbc:hive2://feng01:10000" user="jifeng" password="jifeng"></writeHost> </dataHost> --> 71 ? 72 <!-- <dataHost name="jdbchost" maxCon="1000" minCon="10" balance="0" dbType="mysql" 73 dbDriver="jdbc"> <heartbeat>select user()</heartbeat> <writeHost host="hostM1" 74 url="jdbc:mysql://localhost:3306" user="root" password="123456"> </writeHost> 75 </dataHost> --> 76 </mycat:schema>
9.rule.xml详细描述表的分片规则,格式如下:
1 <tableRule name="分片规则名"> 2 <rule> 3 <columns>分片的列</columns> 4 <algorithm>分片算法名</algorithm> 5 </rule> 6 </tableRule> 7 <function name="分片算法名" class="算法实现类"> 8 <property name="算法参数">参数值</property> 9 </function>
原版样例:
1 <?xml version="1.0" encoding="UTF-8"?> 2 <!DOCTYPE mycat:rule SYSTEM "rule.dtd"> 3 <mycat:rule xmlns:mycat="http://io.mycat/"> 4 <tableRule name="rule1"> 5 <rule> 6 <columns>id</columns> 7 <algorithm>func1</algorithm> 8 </rule> 9 </tableRule> 10 <tableRule name="rule2"> 11 <rule> 12 <columns>user_id</columns> 13 <algorithm>func1</algorithm> 14 </rule> 15 </tableRule> 16 <tableRule name="sharding-by-intfile"> 17 <rule> 18 <columns>sharding_id</columns> 19 <algorithm>hash-int</algorithm> 20 </rule> 21 </tableRule> 22 <tableRule name="auto-sharding-long"> 23 <rule> 24 <columns>id</columns> 25 <algorithm>rang-long</algorithm> 26 </rule> 27 </tableRule> 28 <tableRule name="mod-long"> 29 <rule> 30 <columns>id</columns> 31 <algorithm>mod-long</algorithm> 32 </rule> 33 </tableRule> 34 <tableRule name="sharding-by-murmur"> 35 <rule> 36 <columns>id</columns> 37 <algorithm>murmur</algorithm> 38 </rule> 39 </tableRule> 40 <tableRule name="crc32slot"> 41 <rule> 42 <columns>id</columns> 43 <algorithm>crc32slot</algorithm> 44 </rule> 45 </tableRule> 46 <tableRule name="sharding-by-month"> 47 <rule> 48 <columns>create_time</columns> 49 <algorithm>partbymonth</algorithm> 50 </rule> 51 </tableRule> 52 <tableRule name="latest-month-calldate"> 53 <rule> 54 <columns>calldate</columns> 55 <algorithm>latestMonth</algorithm> 56 </rule> 57 </tableRule> 58 <tableRule name="auto-sharding-rang-mod"> 59 <rule> 60 <columns>id</columns> 61 <algorithm>rang-mod</algorithm> 62 </rule> 63 </tableRule> 64 <tableRule name="jch"> 65 <rule> 66 <columns>id</columns> 67 <algorithm>jump-consistent-hash</algorithm> 68 </rule> 69 </tableRule> 70 <function name="murmur" 71 class="io.mycat.route.function.PartitionByMurmurHash"> 72 <property name="seed">0</property><!-- 默认是0 --> 73 <property name="count">2</property><!-- 要分片的数据库节点数量,必须指定,否则没法分片 --> 74 <property name="virtualBucketTimes">160</property><!-- 一个实际的数据库节点被映射为这么多虚拟节点,默认是160倍,也就是虚拟节点数是物理节点数的160倍 --> 75 <!-- <property name="weightMapFile">weightMapFile</property> 节点的权重,没有指定权重的节点默认是1。以properties文件的格式填写,以从0开始到count-1的整数值也就是节点索引为key,以节点权重值为值。所有权重值必须是正整数,否则以1代替 --> 76 <!-- <property name="bucketMapPath">/etc/mycat/bucketMapPath</property> 77 用于测试时观察各物理节点与虚拟节点的分布情况,如果指定了这个属性,会把虚拟节点的murmur hash值与物理节点的映射按行输出到这个文件,没有默认值,如果不指定,就不会输出任何东西 --> 78 </function> 79 <function name="crc32slot" 80 class="io.mycat.route.function.PartitionByCRC32PreSlot"> 81 </function> 82 <function name="hash-int" 83 class="io.mycat.route.function.PartitionByFileMap"> 84 <property name="mapFile">partition-hash-int.txt</property> 85 </function> 86 <function name="rang-long" 87 class="io.mycat.route.function.AutoPartitionByLong"> 88 <property name="mapFile">autopartition-long.txt</property> 89 </function> 90 <function name="mod-long" class="io.mycat.route.function.PartitionByMod"> 91 <!-- how many data nodes --> 92 <property name="count">3</property> 93 </function> 94 ? 95 <function name="func1" class="io.mycat.route.function.PartitionByLong"> 96 <property name="partitionCount">8</property> 97 <property name="partitionLength">128</property> 98 </function> 99 <function name="latestMonth" 100 class="io.mycat.route.function.LatestMonthPartion"> 101 <property name="splitOneDay">24</property> 102 </function> 103 <function name="partbymonth" 104 class="io.mycat.route.function.PartitionByMonth"> 105 <property name="dateFormat">yyyy-MM-dd</property> 106 <property name="sBeginDate">2015-01-01</property> 107 </function> 108 <function name="rang-mod" class="io.mycat.route.function.PartitionByRangeMod"> 109 <property name="mapFile">partition-range-mod.txt</property> 110 </function> 111 <function name="jump-consistent-hash" class="io.mycat.route.function.PartitionByJumpConsistentHash"> 112 <property name="totalBuckets">3</property> 113 </function> 114 </mycat:rule>
部分常用的分片规则算法说明:
- PartitionByMurmurHash(一致性hash):将物理节点虚拟并映射为一个“一致性hash环”;
- PartitionByCRC32PreSlot(crc32slot 算法):crc32(key)%102400=slot,slot 按照范围均匀分布在 dataNode 上;
- LatestMonthPartion(单月小时拆分):单月内按照小时拆分,最小粒度是小时,可以一天最多 24 个分片,最少 1 个分片,一个月完后下月 从头开始循环;
- PartitionByMonth(自然月):按自然月分片;
- PartitionByRangeMod(范围求模):先进行范围分片计算出分片组,组内再求模;
- PartitionByJumpConsistentHash(一致性hash):另一种一致性hash算法;
- PartitionByFileMap(枚举):通过在配置文件中配置可能的枚举 id,自己配置分片,本规则适用于特定的场景,比如有些业务需要按照省份或区县来做保存,而全国省份区县是固定的;
- PartitionByLong(固定分片 hash 算法):取 id 的二进制低 10 位取模运算,即( id 二进制) &1111111111,partitionCount分片个数,partitionLength分片长度,默认这两个参数的向量积为1024;
- AutoPartitionByLong(范围约定):按照提前规划好分片字段范围计算属于哪个分片,start <= range <= end;
- PartitionByMod(求模):即根据 id 进行十进制求模预算,相比固定分片 hash,此种在批量插入时可能存在批量插入单事务插入多数据分片,增大事务一致性难度;
- PartitionByDate(按天分片):即根据指定的格式,起止日期,按日期划分,如果配置了 sEndDate 则代表数据达到了这个日期的分片后后循环从开始分片插入;
10.情况一:如果DB是一主一从
:需注意这里的主从复制由Mysql实现,Mycat不负责数据复制功能
。只需配置server.xml和schema.xml即可: 本次server.xml实例:
1 <?xml version="1.0" encoding="UTF-8"?> 2 <!DOCTYPE mycat:server SYSTEM "server.dtd"> 3 <mycat:server xmlns:mycat="http://io.mycat/"> 4 <system> 5 <property name="nonePasswordLogin">0</property> 6 <property name="useHandshakeV10">1</property> 7 <property name="useSqlStat">0</property> 8 <property name="useGlobleTableCheck">0</property> 9 <property name="sqlExecuteTimeout">300</property> 10 <property name="sequnceHandlerType">2</property> 11 <property name="sequnceHandlerPattern">(?:(\s*next\s+value\s+for\s*MYCATSEQ_(\w+))(,|\)|\s)*)+</property> 12 <property name="subqueryRelationshipCheck">false</property> 13 <property name="processorBufferPoolType">0</property> 14 <property name="handleDistributedTransactions">0</property> 15 <property name="useOffHeapForMerge">0</property> 16 <property name="memoryPageSize">64k</property> 17 <property name="spillsFileBufferSize">1k</property> 18 <property name="useStreamOutput">0</property> 19 <property name="systemReserveMemorySize">384m</property> 20 <property name="useZKSwitch">false</property> 21 <property name="strictTxIsolation">false</property> 22 <property name="useZKSwitch">true</property> 23 </system> 24 <user name="mycat" defaultAccount="true"> 25 <property name="password">12345678</property> 26 <property name="schemas">dubbo_db</property> 27 </user> 28 </mycat:server>
schema.xml示例:
1 <?xml version="1.0"?> 2 <!DOCTYPE mycat:schema SYSTEM "schema.dtd"> 3 <!-- 数据库配置,与server.xml中的数据库对应 --> 4 <mycat:schema xmlns:mycat="http://io.mycat/"> 5 <schema name="dubbo_db" checkSQLschema="true" sqlMaxLimit="100"> 6 <table name="dubbo_delivery" primaryKey="ID" dataNode="dn1"/> 7 <table name="dubbo_finance" primaryKey="ID" dataNode="dn1 "/> 8 <table name="dubbo_item" primaryKey="ID" dataNode="dn1 " /> 9 <table name="dubbo_order" primaryKey="ID" dataNode="dn1"/> 10 <table name="dubbo_order_detail" primaryKey="ID" dataNode="dn1 "/> 11 <table name="dubbo_stock" primaryKey="ID" dataNode="dn1 " /> 12 </schema> 13 <!-- 分片配置 --> 14 <dataNode name="dn1" dataHost="localhost1" database="dubbo_db" /> 15 <!-- 物理数据库配置 --> 16 <dataHost name="localhost1" maxCon="1000" minCon="10" balance="0" 17 writeType="0" dbType="mysql" dbDriver="native" switchType="1" slaveThreshold="100"> 18 <heartbeat>select user()</heartbeat> 19 <writeHost host="hostM1" url="192.168.1.204:3306" user="root" password="[email protected]"> 20 <readHost host="hostS2" url="192.168.1.205:3306" user="root" password="[email protected]" /> 21 </writeHost> 22 </dataHost> 23 </mycat:schema>
11.情况二,即本期目标架构,DB是两主一从:server.xml不变, 本次schema.xml实例:
1 <?xml version="1.0"?> 2 <!DOCTYPE mycat:schema SYSTEM "schema.dtd"> 3 <!-- 数据库配置,与server.xml中的数据库对应 --> 4 <mycat:schema xmlns:mycat="http://io.mycat/"> 5 <schema name="dubbo_db" checkSQLschema="true" sqlMaxLimit="100"> 6 <table name="dubbo_delivery" primaryKey="ID" dataNode="dn1"/> 7 <table name="dubbo_finance" primaryKey="ID" dataNode="dn1,dn2" rule="rule1"/> 8 <table name="dubbo_item" primaryKey="ID" dataNode="dn1,dn2" rule="rule2"/> 9 <table name="dubbo_order" primaryKey="ID" dataNode="dn1,dn2" rule="sharding-by-murmur"/> 10 <table name="dubbo_order_detail" primaryKey="ID" dataNode="dn1,dn2" rule="sharding-by-month"/> 11 <table name="dubbo_stock" primaryKey="ID" dataNode="dn1" /> 12 </schema> 13 <!-- 分片配置 --> 14 <dataNode name="dn1" dataHost="localhost1" database="dubbo_db" /> 15 <dataNode name="dn2" dataHost="localhost2" database="dubbo_db" /> 16 <!-- 物理数据库配置 --> 17 <dataHost name="localhost1" maxCon="1000" minCon="10" balance="0" 18 writeType="0" dbType="mysql" dbDriver="native" switchType="1" slaveThreshold="100"> 19 <heartbeat>select user()</heartbeat> 20 <writeHost host="hostM1" url="192.168.1.204:3306" user="root" password="[email protected]"> 21 <readHost host="hostS2" url="192.168.1.205:3306" user="root" password="[email protected]" /> 22 </writeHost> 23 </dataHost> 24 <dataHost name="localhost2" maxCon="1000" minCon="10" balance="0" 25 writeType="0" dbType="mysql" dbDriver="native" switchType="1" slaveThreshold="100"> 26 <heartbeat>select user()</heartbeat> 27 <writeHost host="hostM2" url="192.168.1.206:3306" user="root" password="[email protected]" /> 28 </dataHost> 29 </mycat:schema>
本次rule.xml实例:只有使用了分片模式时,才需要配置rule规则,这里写了三种rule,其实也没全部用上:
1 <?xml version="1.0" encoding="UTF-8"?> 2 <!DOCTYPE mycat:rule SYSTEM "rule.dtd"> 3 <mycat:rule xmlns:mycat="http://io.mycat/"> 4 <!--规则定义--> 5 <tableRule name="sharding-by-murmur"> 6 <rule> 7 <columns>id</columns> 8 <algorithm>murmur</algorithm> 9 </rule> 10 </tableRule> 11 <!--自定义规则--> 12 <tableRule name="rule1"> 13 <rule> 14 <columns>userr_id</columns> 15 <algorithm>func1</algorithm> 16 </rule> 17 </tableRule> 18 <tableRule name="rule2"> 19 <rule> 20 <columns>id</columns> 21 <algorithm>func2</algorithm> 22 </rule> 23 </tableRule> 24 <tableRule name="sharding-by-month"> 25 <rule> 26 <columns>create_time</columns> 27 <algorithm>partbymonth</algorithm> 28 </rule> 29 </tableRule> 30 <!--规则算法实现--> 31 <function name="murmur" 32 class="io.mycat.route.function.PartitionByMurmurHash"> 33 <property name="seed">0</property><!-- 默认是0 --> 34 <property name="count">2</property><!-- 要分片的数据库节点数量,必须指定,否则没法分片 --> 35 <property name="virtualBucketTimes">160</property><!-- 一个实际的数据库节点被映射为这么多虚拟节点,默认是160倍,也就是虚拟节点数是物理节点数的160倍 --> 36 <!-- <property name="weightMapFile">weightMapFile</property> 节点的权重,没有指定权重的节点默认是1。以properties文件的格式填写,以从0开始到count-1的整数值也就是节点索引为key,以节点权重值为值。所有权重值必须是正整数,否则以1代替 --> 37 <!-- <property name="bucketMapPath">/etc/mycat/bucketMapPath</property> 38 用于测试时观察各物理节点与虚拟节点的分布情况,如果指定了这个属性,会把虚拟节点的murmur hash值与物理节点的映射按行输出到这个文件,没有默认值,如果不指定,就不会输出任何东西 --> 39 </function> 40 <function name="func1" class="io.mycat.route.function.PartitionByLong"> 41 <!--分片数量,partitionCount*partitionLength=1024--> 42 <property name="partitionCount">2</property> 43 <property name="partitionLength">512</property> 44 </function> 45 <function name="func2" class="io.mycat.route.function.PartitionByLong"> 46 <property name="partitionCount">8</property> 47 <property name="partitionLength">128</property> 48 </function> 49 <function name="partbymonth" 50 class="io.mycat.route.function.PartitionByMonth"> 51 <property name="dateFormat">yyyy-MM-dd</property> 52 <property name="sBeginDate">2015-01-01</property> 53 </function> 54 </mycat:rule>
12.测试:配置好mycat/conf/下的3个xml文件,即配置好了Mycat与物理DB的连接,应用端连接仅需修改连接串端口为Mycat的IP+端口,账号为server.xml中user信息,注意:
要写上默认schema,否则启动应用报Mycat no chose
错,
13.此处有坑!如果Mysql是独立安装在linux上,需要对远程访问打开,否则访问默认仅限本地,导致远程连接一直报错,以开放root
用户远程连接为例:
mysql> use mysql; mysql> update user set Host=‘%‘ where User=‘root‘; mysql> quit;
再重启mysql:
[[email protected] ~]# systemctl restart mysqld
启动Mycat:
[[email protected] ~]# cd /usr/mycat/mycat/bin [[email protected] bin]# ./mycat start
mycat启动成功:
如果启动有问题,使用以下命令查看log:
[[email protected] conf]# tail -F /usr/mycat/mycat/logs/wrapper.log [[email protected] conf]# tail -F /usr/mycat/mycat/logs/mycat.log
然后可以在window上使用如MysqlWorkbench,Navicat测试下是否连接正常,并测试下Mycat连接:
为了集中测试代码,我只改写了finance模块,写个service方法:com.biao.mall.service.DubboFinanceServiceImpl中:
1 //插入1000条数据,看data分布 2 @Override 3 public void testMycat(){ 4 DubboFinanceEntity financeEntity = new DubboFinanceEntity(); 5 for (int i = 0; i < 1000; i++) { 6 financeEntity.setUserId(String.valueOf(i+100)); 7 financeDao.insert(financeEntity); 8 } 9 return "testMycat successfully"; 10 }
写个controller方法跑一跑:
@RestController @RequestMapping("/finance") public class DubboFinanceController { private DubboFinanceServiceImpl financeService; @Autowired public DubboFinanceController(DubboFinanceServiceImpl financeService) { this.financeService = financeService; } @RequestMapping("/mycat") public void testMycat(){ return financeService.testMycat(); } }
启动:ZK---> business --> finance, URI来一个!
DB情况,请看数量和ID分布,红色数字是IP:
这里只测试了两主一从和一种分片规则,其他请君自测!
13.代码地址:其中的day16,https://github.com/xiexiaobiao/dubbo-project.git
后记:
1.认识Mycat的关键特性:
- 支持Mysql原生协议,跨语言,跨平台,跨数据库的通用中间件代理;
- 基于心跳的自动故障切换,支持读写分离,支持MySQL主从;
- 基于Nio实现,有效管理线程,高并发问题;
- 支持数据的多片自动路由与聚合,支持sum,count,max等常用的聚合函数,支持跨库分页;
- 支持通过全局表,ER关系的分片策略,实现了高效的单库多表join查询;
- 支持多租户方案,即同DB下多schema模式;
- 支持全局序列号,解决分布式下的主键生成问题;
- 分片规则丰富,插件化开发,易于扩展,可自定义;
- 集群基于ZooKeeper管理,在线升级,扩容,智能优化,大数据处理(V2.0dev);
- 引入Mycat的无痕切换,我觉得这是最大的优势;
2.认清Mycat的局限性:
- 目前只支持跨库join2个表,不支持3 表及其以上跨库 join ;
- Mycat并没有根据二阶段提交协议实现 XA事务,而是只保证 prepare 阶段数据一致性的弱XA事务,分布式事务场景下,强一致性无法保证;
- 分页排序场景下,会一次查询所有分片,再集中排序分页,有性能瓶颈;
- 不同类型DB适配一般,如Oracle/SQLServer等,由于SQL语法差异,须做彻底的语句兼容测试;
- 没有API配置方法,只有XML方式配置,十分过时;
3.Mycat作为DB上一层的重量级中间件,统一了入口,实际上也破坏了分布式的定义,未能充分发挥DB层的效能,所以也有很多不看好的声音,DB独立使用,更能发挥灵活自由配置,直接对接应用层更为高效。
4.总结:Mycat框架的使用,需持谨慎态度,至少目前来看如此。
推荐阅读:
- Dubbo学习系列之十二(Quartz任务调度)
- Linux下Mysql集群使用
- Dubbo学习系列之十一(Dashboard+Nacos规则推送)
- Dubbo学习系列之十(Sentinel之限流与降级)
- Dubbo学习系列之九(Shiro+JWT权限管理)
原文地址:https://www.cnblogs.com/xxbiao/p/11609574.html