代码拉取完成，页面将自动刷新
smart eredis based on eredis, and add external algorithm for redis sharding,
currently, we provide two algorithm: ketama and random.
The following smart_eredis.config has two pool, the pool_1 are
using ketama algorithm to sharding the keys, and there has two working node,
and one debug node, the pool_2 are using random algorithm you can random to
access every redis server, it is good scenario for read operation with
redis(slave), and it can load balance the redis-server's load.
working node: normally using redis node.
debug node : record every key accessed frequency.
e.g. if you want to access pool_1 to set a key, then the algorithm will help you to find the real redis server, then the key will store in the redis node, if you enabled the debug mode, this key will also store in debug redis node, but the value is the key
accessed countand thelatest accessed time
{smart_eredis,
[{pools,
[
%% {pool_name :: atom(),
%% servers :: list(),
%% scheduling_algorithm :: list(),
%% options :: list()}
{pool_1,[
[{id, 1},
{host, "127.0.0.1"},
{port, 6379},
{database, 1},
{password, ""}],
[{id, 2},
{host, "127.0.0.1"},
{port, 6380},
{database, 1},
{password, ""}],
[{id, debug},
{host, "127.0.0.1"},
{port, 6380},
{database, 0},
{password, ""}]
],
%% [ALGORITHM_MODULE, INIT_OPTIONS, RUNTIME_OPTIONS]
%% init option: while load this pool, it will use
%% this option to initial ketama algorithm
%% runtime option: this parameter are using in every
%% request, see example of random
[ {algorithm, algo_ketama},
{init_options,
[{ring_opt, [{expand_node, false},
{match_operator, '>'},
{concat_char, ":"},
{copies_gen_type, specific}]},
{nodes, [{1, "redis:node_001:6379", 100, 1},
{2, "redis:node_001:6380", 100, 2}]}]},
{runtime_options, []}
],
[{debug, true}]
},
{pool_2,[
[{id, 1},
{host, "127.0.0.1"},
{port, 6379},
{database, 1},
{password, ""}],
[{id, 2},
{host, "127.0.0.1"},
{port, 6380},
{database, 1},
{password, ""}],
[{id, debug},
{host, "127.0.0.1"},
{port, 6380},
{database, 0},
{password, ""}]
],
[
{algorithm, algo_random},
{init_options, []},
{runtime_options, [{ids, [1,2]}]}
],
[{debug, true}]
}
]}
]
}.
tiny different of eredis, it is not necessary to use Client(pid), and you need show clearly the key in the scecond parameter
eredis
{ok, <<"OK">>} = eredis:q(C, ["SET", "foo", "bar"]).
smart_eredis
{ok, <<"OK">>} = smart_eredis:q(pool_1, "foo", ["SET", "foo", "bar"]).
you need to implament following behavior
-module(smart_eredis_algorithm).
-callback init( PoolName ::atom(),
InitOption :: list()) ->
ok | {error, Reason :: term()}.
-callback get_client_id( PoolName :: atom(),
Key :: string(),
Options :: list()) ->
ok | {error, Reason :: term()}.
example:
-module(algo_random).
-behaviour(smart_eredis_algorithm).
-export([init/2, get_client_id/3]).
init(_PoolName, _Options)->
ok.
%% {algorithm, algo_random},
%% {init_options, []},
%% {runtime_options, [{ids, [1,2]}]}
get_client_id(_PoolName, _Key, Options)->
Ids = proplists:get_value(ids, Options, []),
Len = length(Ids),
case Len > 0 of
true ->
random:seed(os:timestamp()),
{ok, random:uniform(Len)};
false ->
{error, no_random_ids}
end.
此处可能存在不合适展示的内容，页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容，可点击提交进行申诉，我们将尽快为您处理。
1. 开源生态
2. 协作、人、软件
3. 评估模型