sharded_eredis

最新推荐文章于 2024-09-14 19:55:29 发布

weixin_34403693

最新推荐文章于 2024-09-14 19:55:29 发布

阅读量61

点赞数

文章标签：数据库 git

原文链接：https://yq.aliyun.com/articles/503984

版权

sharded_eredis is collection of pools of Redis clients, using eredis and poolboy. Each pool points to a different shard.

This project modifies the original repository by including a consistent hashing library to simplify presharding, suggested by antirez. Read the original blogpost here.

The name has been changed so as not to induce confusion for those originally using hiroeorz's sharded_eredis; the functionality has changed significantly.

eredis: https://github.com/wooga/eredis

poolboy: https://github.com/devinus/poolboy

Caveats!!

This library uses Redis but in a distributed fashion. Multi-object operations are no longer guaranteed to be atomic and are not supported by this library. Examples include multisets, multigets, source-destination operations, and multi-object transactions. Transactions technically still work if all keys involved in the transaction are on the same machine but the Redis transaction was not built to be distributed.

Setup

git clone git://github.com/jeremyong/sharded_eredis.git
cd eredis_pool
make get-deps
make

Testing

make test

Config

Add new pools. This configuration specifies 4 shards.

  {env, [
          {global_or_local, local}
          {pools, [
                   {pool0, [
                              {size, 10},
                              {max_overflow, 20}
                              {host, "127.0.0.1"},
                              {port, 6378}
                             ]},
                   {pool1, [
                              {size, 10},
                              {max_overflow, 20},
                              {host, "127.0.0.1"},
                              {port, 6380}
                             ]},
                   {pool2, [
                              {size, 10},
                              {max_overflow, 20},
                              {host, "127.0.0.1"},
                              {port, 6381},
                             ]}
                   {pool3, [
                              {size, 10},
                              {max_overflow, 20},
                              {host, "127.0.0.1"},
                              {port, 6382},
                             ]}
                  ]}
        ]}

You can include this using the erlang -config option.

The global_or_local option is used to determine whether the pools are registered as local or global. Personally, I start this application on every instance and run pools on each instance locally to minimize network latency.

Examples

application start.

 sharded_eredis:start().
 ok

key-value set and get

 sharded_eredis:q(["SET", "foo", "bar"]).
 {ok,<<"OK">>}
 
 sharded_eredis:q(["GET", "foo"]).       
 {ok,<<"bar">>}

Note that the name of the pool does not need to be supplied with each query (in contrast to prior behavior). The library will automatically determine which pool associated to the correct shard should be invoked.

The Redis documentation can be referred to here.

本文作者：陈群

本文来自云栖社区合作伙伴rediscn，了解相关信息可以关注redis.cn网站。