Redis07-Redis single node capacity problems, twemproxy, predixy use

Redis single node capacity problem

A single node capacity issues

We in the actual scene, often encounter a single node capacity issues.

1. service splitter , data classification

2. to the data can not be split when sliced data can be

• Modulo hash (under the influence of a distributed scalable 3%, 4%, if a machine pay more, will receive the impact)
• Random logic (can be put to, but get it out)
  • Solution: two machines simultaneously store a list, then the client directly to connect two redis, carried out with two consumer
• Consistent hashing algorithm
  • crc16 crc32 md5 sha1 sha256
  • No modulo-width 16-bit, 16-bit hash abstract a ring, to calculating hashing algorithm

一致性哈希算法（哈希环）：
1.求出memcached服务器（节点）的哈希值，并将其配置到0～232的圆（continuum）上。
2.采用同样的方法求出存储数据的键的哈希值，并映射到相同的圆上。
3.从数据映射到的位置开始顺时针查找，将数据保存到找到的第一个服务器上。如果超过232仍然找不到服务器，就会保存到第一台memcached服务器上。
来源：（https://www.cnblogs.com/williamjie/p/9477852.html）

3. advantages and disadvantages

Pros: add node to other nodes can indeed share the pressure (and will not cause a global reshuffle)

Cons: New node will cause a small portion of the data can not hit

二、twemproxy

twemproxy fragmentation agent is a mechanism by open twitter, twemproxy as a proxy can accept multiple access programs, in accordance with the routing rules, the forwarding Redis for the background of each server, then backtrack, the program a good solution Redis examples carrying capacity.

installation

git clone https://github.com/twitter/twemproxy.git
如果报错，执行：yum update nss
yum install automake libtool
autoreconf -fvi
如果报错，执行
wget -O /etc/yum.repos.d/epel.repo http://mirrors.aliyun.com/repo/epel-6.repo
yum clean all
yum install autoconf268.noarch -y
autoreconf268 -fvi
./configure --enable-debug=full
make
查看服务文件
cd scripts
nutcracker.init
拷贝这个文件进/etc/init.d目录
拷贝编译运行文件进/usr/bin目录
拷贝conf文件夹进/etc/nutcracker目录
进入/etc/nutcracker，修改nutcracker.yml进行配置

alpha:
  listen: 127.0.0.1:22121
  hash: fnv1a_64
  distribution: ketama
  auto_eject_hosts: true
  redis: true
  server_retry_timeout: 2000
  server_failure_limit: 1
  servers:
   - 127.0.0.1:6379:1
   - 127.0.0.1:6380:1

之后开启nutcracker服务，开启service服务，之后连接redis-cli进行连接22121端口
我们通过nutcracker进行get和set，我们在nutcracker不支持的命令：
keys *
watch k1
multi
这些命令都不支持

predixy software, it can also be used as a substitute

wget https://github.com/joyieldInc/predixy/releases/download/1.0.5/predixy-1.0.5-bin-amd64-linux.tar.gz

Modify predixy.conf

打开Bind 127.0.0.1:7617
打开include sentinel.conf

Modify the Sentinel 26379

port 26379
sentinel monitor ooxx 127.0.0.1 36379 2
sentinel monitor xxoo 127.0.0.1 46379 2

Modify the Sentinel 26380

port 26380
sentinel monitor ooxx 127.0.0.1 36379 2
sentinel monitor xxoo 127.0.0.1 46379 2

The following were started omitted.

After the direct test