MongoDB之Replica Set+Sharding架构

MongoDb Replica Set解决了容错和单点故障问题，但单台机器的存储和承受能力有限，Sharding就是为了海量存储和动态扩容而产生的。这才有了Replica Set+Sharding高可用架构。
Sharding Cluster主要包括如下三部分：




Shards:每个shard都是replica set,具有自动备份、容错、恢复能力，当然在开发环境你可配成单个mongod

Config Server:存储metadata,包括每个shard的基本信息和chunk信息，生产环境至少有3个config server

Mongos:集群的入口，客户端与之交互，复杂把请求转发到对应的shard上。

搭建MongoDB Replica Set+Sharding环境：
拓扑图：



说明：由于本地测试，部署将在一台机器完成，并且mongos和config server都只用了一个实例。
配置：
副本集rep1配置：

#master.conf配置
dbpath=/data/mongo_cluster/master
logpath=/data/mongo_cluster/master/logs/master.log
pidfilepath=/data/mongo_cluster/master/master.pid
logappend=true          
replSet=rep1
bind_ip=192.168.15.130
port=10000
fork=true
journal=true
shardsvr=true

#slave.conf配置
dbpath=/data/mongo_cluster/slave
logpath=/data/mongo_cluster/slave/logs/slave.log
pidfilepath=/data/mongo_cluster/slave/slave.pid
logappend=true
replSet=rep1
bind_ip=192.168.15.130
port=10001
fork=true
journal=true
shardsvr=true

#arbiter.conf配置
dbpath=/data/mongo_cluster/arbiter
logpath=/data/mongo_cluster/arbiter/logs/arbiter.log
pidfilepath=/data/mongo_cluster/arbiter/arbiter.pid
logappend=true
replSet=rep1
bind_ip=192.168.15.130
port=10002
fork=true
journal=true
shardsvr=true

副本集rep2配置：

#master.conf配置
dbpath=/data/mongo_cluster2/master
logpath=/data/mongo_cluster2/master/logs/master.log
pidfilepath=/data/mongo_cluster2/master/master.pid
logappend=true          
replSet=rep2
bind_ip=192.168.15.130
port=10004
fork=true
journal=true
shardsvr=true

#slave.conf配置
dbpath=/data/mongo_cluster2/slave
logpath=/data/mongo_cluster2/slave/logs/slave.log
pidfilepath=/data/mongo_cluster2/slave/slave.pid
logappend=true
replSet=rep2
bind_ip=192.168.15.130
port=10005
fork=true
journal=true
shardsvr=true

#arbiter.conf配置
dbpath=/data/mongo_cluster2/arbiter
logpath=/data/mongo_cluster2/arbiter/logs/arbiter.log
pidfilepath=/data/mongo_cluster2/arbiter/arbiter.pid
logappend=true
replSet=rep2
bind_ip=192.168.15.130
port=10006
fork=true
journal=true
shardsvr=true

配置Replica Set主、备、仲裁节点:(点击查看)
config server配置：

#config server配置configsvr.conf
dbpath=/data/mongo_config_server
logpath=/data/mongo_config_server/logs/configsvr.log
pidfilepath=/data/mongo_config_server/configsvr.pid
logappend=true          
bind_ip=192.168.15.130
port=10007
fork=true
journal=true
configsvr=true

mongos配置：

#mongos配置 mongos.conf
logpath=/data/mongos/logs/mongos.log
pidfilepath=/data/mongos/mongos.pid
logappend=true          
bind_ip=192.168.15.130
port=10008
fork=true
configdb=192.168.15.130:10007

启动：

#启动shard
/usr/local/mongodb3.0.5/bin/mongod -f /data/mongo_cluster/master/master.conf
/usr/local/mongodb3.0.5/bin/mongod -f /data/mongo_cluster/slave/slave.conf
/usr/local/mongodb3.0.5/bin/mongod -f /data/mongo_cluster/arbiter/arbiter.conf

/usr/local/mongodb3.0.5/bin/mongod -f /data/mongo_cluster2/master/master.conf
/usr/local/mongodb3.0.5/bin/mongod -f /data/mongo_cluster2/slave/slave.conf
/usr/local/mongodb3.0.5/bin/mongod -f /data/mongo_cluster2/arbiter/arbiter.conf

#启动config server
/usr/local/mongodb3.0.5/bin/mongod -f /data/mongo_config_server/configsvr.conf 

#启动mongos
/usr/local/mongodb3.0.5/bin/mongos -f /data/mongos/mongos.conf

配置mongos，设置分片规则：
(1)连接mongos与replica set

#进入mongos
/usr/local/mongodb3.0.5/bin/mongo 192.168.15.130:10008
#连接mongos与replica set
mongos> db.runCommand( { addshard : "rep1/192.168.15.130:10001"});
{ "shardAdded" : "rep1", "ok" : 1 }
mongos> db.runCommand( { addshard : "rep2/192.168.15.130:10004"});
{ "shardAdded" : "rep2", "ok" : 1 }
#查看分片配置
mongos> db.runCommand( { listshards : 1 } );

(2)设置分片规则并测试

#指定testdb分片生效
db.runCommand( { enablesharding :"testdb"});
#指定数据库里需要分片的集合和片键
db.runCommand( { shardcollection : "testdb.table1",key : {id: 1} } )

#测试
> use  testdb; 
#插入测试数据
> for (var i = 1; i <= 1000; i++) db.table1.save({id:i,"test1":"testval1"});
#查看分片情况如下
> db.table1.stats();

参考：

http://docs.mongodb.org/v3.0/core/sharding-introduction/

MongoDB Sharding机制