libchan: like Go channels over the network
Libchan is an ultra-lightweight networking library which lets network services communicate in the same way that goroutines communicate using channels:
- Simple message passing
- Synchronization for concurrent programming
- Nesting: channels can send channels
Libchan supports the following transports out of the box:
- In-memory Go channel
- Unix socket
- Raw TCP
- TLS
- HTTP2/SPDY
- Websocket
This provides great flexibility in scaling an application by breaking it down into loosely coupled concurrent services. The same application could be composed of goroutines communicating over in-memory channels; then transition to separate unix processes, each assigned to a processor core, and communicating over high-performance IPC; then to a cluster of machines communicating over authenticated TLS sessions. All along it benefits from the concurrency model which has made Go so popular.
Not all transports have the same semantics. In-memory Go channels guarantee exactly-once delivery; TCP, TLS, and the various HTTP socket families do not guarantee delivery. Messages arrive in order but may be arbitrarily delayed or lost. There are no ordering invariants across channels.
An explicit goal of libchan is simplicity of implementation and clarity of spec. Porting it to any language should be as effortless as humanly possible.
Focused on not reinventing the wheel
Because remote libchan sessions are regular HTTP2 over TLS sessions, they can be used in combination with any standard proxy or authentication middleware. This means libchan, when configured properly, can be safely exposed on the public Internet. It can also be embedded in an existing rest API using an http1 and websocket fallback.
How is it different from RPC or REST?
Modern micro-services are not a great fit for classical RPC or REST protocols because they often rely heavily on events, bi-directional communication, stream multiplexing, and some form of data synchronization. Sometimes these services have a component which requires raw socket access, either for performance (file transfer, event firehose, database access) or simply because they have their own protocol (dns, smtp, sql, ssh, zeromq, etc). These components typically need a separate set of tools because they are outside the scope of the REST and RPC tools. If there is also a websocket or ServerEvents transport, those require yet another layer of tools.
Instead of a clunky patchwork of tools, libchan implements in a single minimalistic library all the primitives needed by modern micro-services:
-
Request/response with arbitrary structured data
-
Asynchronous events flowing in real-time in both directions
-
Requests and responses can flow in any direction, and can be arbitrarily nested, for example to implement a self-registering worker model
-
Any message serialization format can be plugged in: json, msgpack, xml, protobuf.
-
Raw file descriptors can be "attached" to any message, and passed under the hood using the best method available to each transport. The Go channel transport just passes os.File pointers around. The unix socket transport uses fd passing which makes it suitable for high-performance IPC. The tcp transport uses dedicated http2 streams. And as a bonus extension, a built-in tcp gateway can be used to proxy raw network sockets without extra overhead. That means libchan services can be used as smart gateways to a sql database, ssh or file transfer service, with unified auth, discovery and tooling and without performance penalty.
Example usage
Here's an example implementing basic RPC-style request/response. We gloss over error handling to tersely demonstrate the core concepts.
On the client:
var ch libchan.Sender
// Send a message, indicate that we want a return channel to be automatically created
ret1, err := ch.Send(&libchan.Message{Data: []byte("request 1!"), Ret: libchan.RetPipe})
// Send another message on the same channel
ret2, err := ch.Send(&libchan.Message{Data: []byte("request 2!"), Ret: libchan.RetPipe})
// Wait for an answer from the first request. Set flags to zero
// to indicate we don't want a nested return channel.
msg, err := ret1.Receive(0)
On the server:
var ch libchan.Receiver
// Wait for messages in a loop
// Set the return channel flag to indicate that we
// want to receive nested channels (if any).
// Note: we don't send a nested return channel, but we could.
for {
msg, err := ch.Receive(libchan.Ret)
msg.Ret.Send(&libchan.Message{Data: []byte("this is an extremely useful response")});
}
Creators
Solomon Hykes
Additional Implementations
Java
Javascript / Node.js
Copyright and license
Code and documentation copyright 2013-2014 Docker, inc. Code released under the Apache 2.0 license. Docs released under Creative commons.
