BrB - Easy and Fast distributed ruby
BrB is a simple, fully transparent and extremely fast interface for doing simple distributed Ruby. The core of the architecture is provided by EventMachine (a fast and reliable IO event library).
BrB was built in order to achieve these 4 main goals :
-
Simple and fast message passing between distant Ruby processes.
-
Message passing with of return values when needed.
-
Being extremely fast in order to handle more than a few thousand messages per second.
-
Being completely transparent for developer.
The principle is simple and inspired from Drb (standard distributed ruby library) : A process exposes an object over the network and any ruby process (after having established a connection tunnel) can directly call a method on the exposed object. BrB handles that part, so it’s fully transparent in the Ruby code.
BrB only support message passing with Marshable dumpable object : String, symbol, Array, hash, Number, Object etc… That mean you can not send file descriptor, Thread or another funky things like that :)
For any question, use the Ruby BrB google group: groups.google.com/group/ruby-brb
Main Functionalities
-
Unlimited objects exposed
-
Processes can expose object and be client to exposed object too at the same time
-
Do not wait for return by default : just do simple message passing
-
Handle return values without blocking with the usage of a simple block
-
Blocking wait for a return value if needed by simply adding _block at the end of the method name
-
Transmission of Exception when blocking call
-
Thread safe if used correctly with Event Machine
How it works
First of all, a process declare himself as a sever and expose any object on a given address and port. Then, any number of distant processes can create a Tunnel with that server and can expose an object in exchange too. After connection are ready, just call method on the tunnel. It will just act like normal method calling on the exposed object !
What BrB is designed for ?
-
Doing Simple message passing between ruby process.
-
Connecting hundred of ruby process transparently.
-
Building a real-time scalable (game) server
-
Taking important load on a server easily just by distributing the load on multiple BrB instance.
-
Taking advantage of multi-core and multi-threaded systems.
TODO
-
Writing more examples
-
Publish Benchmarks VS drb
-
Improve logging mechanism
-
Clean up
Contributors
-
kwi (Guillaume Luccisano)
-
bwalton (Brian Walton)
-
dpree (Jens Bissinger)
Copyright © 2009-2010 Guillaume Luccisano - g-mai|: guillaume.luccisano, released under the MIT license