Last Modified February 15 2019
What is Python Banyan?
Python Banyan is a lightweight, reactive framework used to create flexible, non-blocking, event driven, asynchronous applications. It was designed primarily to aid in the implementation of real-time physical computing applications for devices such as the Raspberry Pi, ESP8266, and Arduino, but may easily be applied to projects outside of the physical programming domain.
Banyan uses a publish/subscribe model that is similar to MQTT, but is much faster than MQTT in real-time control and data monitoring applications. A benchmark comparison of Banyan and MQTT is provided later in this guide to illustrate Banyan's efficiency. If your application requires connectivity to an MQTT broker, a Gateway is provided.
Most traditional physical computing libraries or frameworks use an object oriented model, that yield a single, tightly coupled, monolithic executable image. Python Banyan uses an extension of the object oriented model, called the component based service oriented architectural model. A component based application consists of a set of independent, loosely coupled components, that communicate with one another through a set of platform independent, user defined, protocol messages transmitted over a TCP network.
Banyan uses a central message distribution mechanism called the Backplane. When first starting a Banyan application, the Backplane is brought on-line, followed by the application's components. At startup, each component automatically "plugs" itself into the Backplane establishing a TCP/IP communication channel between itself and the Backplane.
After communication is established, components may publish messages to the Backplane, receive messages from other components or both publish and receive messages. A major feature of Python Banyan is that it has the capability to incorporate Numpy data into a protocol message seamlessly and easily.
Without modification, a Banyan application's components can run on a single computer or distributed across multiple computers. This will be demonstrated later on in this document.
A Little More Detail About Python Banyan
Banyan Base Classes
All Banyan components are created by inheriting from one of the supplied Banyan base classes. The base class methods are overridden as needed to support the component's specific needs.
The base classes act as a wrapper for the ZeroMQ embeddable networking library that has been configured to operate in a Publish/Subscribe networking pattern. In addition, the base class encodes and decodes protocol messages for efficient network transmission, using MessagePack.
Banyan Protocol Messages
Banyan protocol messages consist of 2 parts, a "topic" and a "payload".
The Topic String
A message topic consists of a simple string. To receive messages, a component subscribes to one or more topics. A subscribing component receives all messages published containing the desired topic.
When a message is received, the topic is compared to the subscribed topic, and if it matches, the message is placed on the component's receive queue. Any topic that begins with the subscribed topic string is considered a match even if the received message topic contains additional text. For example, if you subscribe to the topic "abcd", and a message is received with a topic of "abcde", the "abcde" message will be placed on the receive queue. Any messages that have topics that begin with the subscribed topic string are considered a match even though the topic may contain additional text.
The Message Payload
Message payloads consist of a Python dictionary type structure, containing one or more name/value items. By utilizing a dictionary, message data can be quickly de-referenced and processed.
Where Is Python Banyan Being Used?
Commercially, Python Banyan has been chosen by Palace Games in San Francisco to monitor and control their Palace Games Edison Room. The Edison room contains several hundred sensors and actuators, and Banyan helps tie all the devices and their associated micro-controllers together, allowing them to communicate quickly and transparently with one another. Banyan supports data transfers of Numpy data, which the Edison room uses to efficiently model their data. You can read about Palace Games on the Raspberry Pi Blog.
Banyan is also being used in numerous physical computing projects utilizing the Raspberry Pi, Arduino, and ESP8266. An earlier version of Python Banyan, called razmq, was used to control a Raspberry Pi robot.
Banyan Application Design
When designing a Banyan Component, it is good practice to keep it as small as possible. By limiting a component to a small area of concern, testing is greatly simplified and the potential for component reuse increases.
Some Design Guidelines
A component should subscribe to the minimal number of message topics as possible.
Keep messages as short as possible.
Limit the component to its task at hand. For example if you wish to log application activity, create a specific logging component. All other components can take advantage of this logging facility in a consistent fashion. It also offloads logging I/O from the main component, promoting higher performance for the system as a whole.