eharbor is an innovative Erlang library designed to enhance the resilience and efficiency of distributed systems under heavy load. Unlike traditional approaches that rely solely on queues, eharbor intelligently combines load-shedding, request deduplication, and concurrency control to provide a more robust solution to system overload.

• Intelligent Load-Shedding: Goes beyond simple queuing to proactively manage system load, preventing overload before it occurs.
• Request Deduplication: Efficiently handles multiple identical requests by processing a single instance and sharing the result, significantly reducing redundant workload.
• Concurrency Management: Limits the number of concurrent operations, protecting backend systems (like databases) from being overwhelmed.
• Configurable Buffering: Offers fine-tuned control over buffer sizes and system limits, ensuring optimal performance tailored to specific operational needs.

Traditional overload protection mechanisms often rely heavily on queuing requests. However, queues alone cannot prevent system overload; they merely postpone the problem. eharbor addresses this by implementing an intelligent system that manages queued requests, actively reduces unnecessary workload, and controls the concurrency of task processing.

To install eharbor, add it to your rebar.config dependencies:

{deps, [
    {eharbor, "1.0.0"}
]}.

Then run rebar3 to fetch and compile the library:

$ rebar3 compile

First, set up your eharbor configuration:

%% Sample configuration
Config = #{
  name => my_harbor,
  backlog => 500,  %% Buffer size
  piers => 50,     %% Concurrency limit
  breakwater_limit => 1000,  %% Max waiting processes
  dedup => true,   %% Enable request deduplication
  ordered => true  %% Maintain operation order
}.

Start an instance of eharbor with your configuration or add it to your supervisor tree:

{ok, _Pid} = eharbor:start_link(Config).

Use eharbor to run tasks efficiently:

%% Execute the task through `eharbor`
Result = eharbor:run(fun some_mod:slow_function/1, [Param], Config).

%% Start eharbor
{ok, _Pid} = eharbor:start_link(Config).

%% Run the function via eharbor
Result = eharbor:run(fun some_mod:slow_function/1, [Arg], Config).

Some parameters sometimes need to be excluded from deduplication matching.

%% Start eharbor
{ok, _Pid} = eharbor:start_link(Config).

%% Define a resource-intensive function
SourceIP = ...
IntensiveFun = fun(Arg) -> some_mod:heavy_computation(Arg, SourceIP) end.

%% Run the function via eharbor
Result = eharbor:run(IntensiveFun, [Arg], Config).

For detailed documentation, including advanced configurations and operational guidelines, please refer to our repository's doc directory or visit our online documentation.

Contributions to eharbor are welcome! Please feel free to contribute, whether it's feature requests, bug reports, or code contributions. For code contributions, please fork the repository and submit a pull request with your changes.

eharbor is released under the Apache License 2.0. Please look at the LICENSE file for more details.