Home

Awesome

Diameter Base Protocol

Go Report Card Test Status Go Reference Latest

Package go-diameter is an implementation of the Diameter Base Protocol RFC 6733 and a stack for the Go programming language.

Status

The current implementation is solid and works fine for general purpose clients and servers. It can send and receive messages efficiently as well as build and parse AVPs based on dictionaries.

See the API documentation at https://pkg.go.dev/github.com/fiorix/go-diameter/v4

Features

Getting started

The easiest way to get started is by trying out the client and server example programs.

With Go 1.11 and newer (preferred), you can start the client and server already:

export GO111MODULE=on
go run github.com/fiorix/go-diameter/v4/examples/server
go run github.com/fiorix/go-diameter/v4/examples/client -hello

Without modules, use standard procedure:

go get github.com/fiorix/go-diameter/examples/...
go run github.com/fiorix/go-diameter/examples/server
go run github.com/fiorix/go-diameter/examples/client -hello

Source code is your best friend. Check out other examples and test cases.

Performance

Clients and servers written with the go-diameter package can be quite performant if done well. Besides Go benchmarks, the package ships with a simple benchmark tool to help testing servers and identifying bottlenecks.

In the examples directory, the server has a pprof (http server) that allows the go pprof tool to profile the server in real time. The client can perform benchmarks using the -bench command line flag.

For better performance, avoid logging diameter messages. Although logging is very useful for debugging purposes, it kills performance due to a number of conversions to make messages look pretty. If you run benchmarks on the example server, make sure to use the -s (silent) command line switch.

TLS degrades performance a bit, as well as reflection (Unmarshal). Those are important trade offs you might have to consider.

Besides this, the source code (and sub-packages) have function benchmarks that can help you understand what's fast and isn't. You will see that parsing messages is much slower than writing them, for example. This is because in order to parse messages it makes numerous dictionary lookups for AVP types, to be able to decode them. Encoding messages require less lookups and is generally simpler, thus faster.

Contribute

In case you want to add new AVPs, please add them to diam/dict/testdata xml files. Then regenerate the go models using ./autogen.sh you will find at diam folder. This will modify files at diam/dict to include your changes.

Before submitting PR, please run make test to test your changes. Or do it manually:

	go test ./...

You also have the option to run the test using a Linux VM through Docker (this is not mandatory). To do so, run make test_docker. Runing test on Linux can be useful in case you add sctp tests. Note you will need to install docker and docker-compose.