mysql cluster简介

我半年前的帖子,已经许久没有接触mysql了,而且mysql的cluster当初就没有定型，所以现在放在这里不知过时了没有，仅供参考。

---

最近在考虑整个系统的容灾和备份，由于用的是mysql, 如果mysql有现成的方案就最好了。
google一下，发现原来MySQL 4.1.x版本已经推出了cluster功能，采用NDB技术，从索爱购买的的技术，（什么时候俺的索爱手机能装mysql?)还有一下评论：
MySQL Cluster is a shared-nothing main memory database clustering technology that MySQL bought from a division of Ericsson. It removes that pesky single point of failure from the database portion of your architecture, but it's not a panacea -- if you've got a lot of data, you need a lot of RAM. In a given cluster, you can store (Nodes x Available-RAM-per-node / Redundancy-level) bytes of data.

For example, if you arrange your 10 node! s in redunancy pairs (each piece of data lives in two places, so "Redundancy-level" is 2), and each node has 4Gb of RAM, then your cluster can hold 10 x 4Gb / 2 = about 20Gb of data. It's "about 20Gb" because all 4Gb on each node isn't really available to MySQL Cluster -- the OS needs some RAM, too.

Future versions of MySQL Cluster will use memory as a disk-backed cache so your storage is bounded by available disk instead of available RAM.

NDB应该和InnoDB、MyISAM的使用方式一样, 可以混合使用。既不同的database可以采取不同的方式，下一步要来验证一下。
上文也可以看出，MySQL采用全内存数据库，The recommendation is to have memory 2x the size of your database size + 10%，这样可能速度会比较快，但对内存要求大。但看起来，在使用时，最好是将database全装到内存里。cluster也是类似的机制，数据都是在 内存里，MySQL
Cluste! r is implemented using a Shared Nothing Architecture, so each ! database node has its own disk and
memory storage rather than requiring a shared disk which adds to the expense and can limit performance
in proprietary solutions. MySQL Cluster keeps all data in memory for transaction and fast failover
purposes and limits Input/Output bottlenecks by asynchronously writing transactions logs to disk.
不知道这样会有什么风险，还要进一步试一下。如果比较可靠，就是个非常好的消息。


具体实现(2机cluster):


192.168.0.1服务器(主机)
mkdir /var/lin/mysql-cluster
建立config.ini文件，内容如下
[NDBD DEFAULT]
NoOfReplicas= 2

[MYSQLD DEFAULT]
[NDB_MGMD DEFAULT]
[TCP DEFAULT]

[NDB_MGMD]
HostName= 192.168.0.1

[NDBD]
HostName= 192.168.0.1
DataDir= /var/lib/mysql-cluster

[NDBD]
HostName= 192.168.0.2
DataDir= /var/lib/mysql-cluster
!
[MYSQLD]
[MYSQLD]
[MYSQLD]

mysql安装目录为/usr/local/mysql
在/usr/local/mysql/data下建立文件my.cnf(从/usr/local/mysql/support-files
下COPY过来),内容如下
[mysqld]
下加入
ndbcluster

最后加入
[mysql_cluster]
ndb-connectstring=192.168.0.1

/////////********************
进入/var/lin/mysql-cluster，执行  //进入Mysql中的Cluster
ndb_mgmd
ndbd --initial
/etc/init.d/mysql restart(重起MYSQL)
/////////********************

192.168.0.2服务器（子机）
mkdir /var/lin/mysql-cluster
建立config.ini文件，内容如下
[NDBD DEFAULT]
NoOfReplicas= 2

[MYSQLD DEFAULT]
#[NDB_MGMD DEFAULT]
[TCP DEFAULT]

[NDB_MGMD]
HostName= 192.168.0.1

[NDBD]
HostName= 192.168.0.1
DataDir= /var/lib/mysql-cluster

[NDBD] HostName= 192.168.0.2
DataDir= /var/lib/mysql-clust! er

[MYSQLD]
[MYSQLD]
[MYSQLD]

mysql安装目录为/usr/local/mysql-max
在/usr/local/mysql/data-max下建立文件my.cnf(从/usr/local/mysql/support-
files下 COPY过来),内容如下
[mysqld]
ndbcluster
ndb-connectstring=192.168.0.1

[ndbd]
connect-string=192.168.0.1

# provide connectstring for management server host (default port: 1186)
[ndb_mgm]
connect-string=192.168.0.1

# provide location of cluster configuration file
[ndb_mgmd]
config-file=/var/lib/mysql-cluster/config.ini

[mysql_cluster]
ndb-connectstring=192.168.0.1

ok 这样，cluster 就构建成功了！ 用客户端访问：192.168.0.1，即可完成自动的负载均衡。