Skip to content

AMQP based microservices communication and orchestration

License

Notifications You must be signed in to change notification settings

imubit/amqp-fabric

Repository files navigation

Built Status PyPI-Server Coveralls

amqp-fabric

AMQP Fabric is an AMQP based microservice orchestration and communication framework.

Description

AMQP Fabric is a very simple microservice communication and orchestration mechanism based on AMQ protocol. Instead of relying on multiple technologies and orchestration frameworks - it's a "one-stop shop" library for implementing a light-weight microservices topology.

Each service in the ecosystem can publish it's own API and use API of another service. A service can send an asynchronous stream of data that other services can subscribe to. Services can optionally send periodic "keep-alives" to allow tracking its uptime.

Features

  • Microservice communication via synchronous API (RPC)
  • Asynchronous data transmission
  • Decentralized registry
  • Remote logging based on standard Python logging mechanism
  • High-availability
  • Secure and firewall friendly access from remote locations

Installation

pip install amqp-fabric

Getting Started

Each service participating in the ecosystem is assigned with:

  • "Domain" - (i.e. project1) any string identifying a group services communicating with each other. Different domains can co-exist under the same AMQP broker.
  • "Service Type" - (i.e. media_encoder) - services holding the same service type, should have the same API.
  • "Service Id" - (i.e. encoder1) Multiple services of the same type can be distinguished by a different Id.
  • "Service Version" - evolution of the services and their API should be tracked by a version

High Availability

Multiple services with the same Domain, Type and Id - will create a high-availability "clique" - API calls will be redirected to the next available service.

Server Side Example

import asyncio
from amqp_fabric.amq_broker_connector import AmqBrokerConnector
from amqp_fabric.abstract_service_api import AbstractServiceApi

# API Definition
class MyServiceApi(AbstractServiceApi):

    def multiply(self, x, y):
        return x * y


class MyService:

    amq = None

    async def init(self):

        self.amq = AmqBrokerConnector(
            amqp_uri="amqp://guest:guest@127.0.0.1/",
            service_domain="my_project",
            service_id="my_app",
            service_type="server_app",
            keep_alive_seconds=5)
        await self.amq.open(timeout=10)

        api = MyServiceApi()
        await self.amq.rpc_register(api)

    async def close(self):
        await self.amq.close()


def run_event_loop():
    agent = MyService()
    loop = asyncio.get_event_loop()
    loop.run_until_complete(agent.init())

    try:
        loop.run_forever()
    except (KeyboardInterrupt, SystemExit):
        pass
    finally:
        loop.run_until_complete(agent.close())
        loop.run_until_complete(loop.shutdown_asyncgens())
        loop.close()

if __name__ == "__main__":

    run_event_loop()

Client Side Example

import asyncio
from amqp_fabric.amq_broker_connector import AmqBrokerConnector


async def exec_multiply():

    amq = AmqBrokerConnector(
        amqp_uri="amqp://guest:guest@127.0.0.1/",
        service_domain="my_project",
        service_id="my_client",
        service_type="client_app",
        keep_alive_seconds=5)
    await amq.open(timeout=10)


    srv_proxy = await amq.rpc_proxy("my_project", "my_app", "server_app")

    result = await srv_proxy.multiply(x=5, y=7)
    print(f'result = {result}')

    await amq.close()


if __name__ == '__main__':

    task = exec_multiply()

    loop = asyncio.get_event_loop()
    loop.run_until_complete(task)

Uptime Tracking

amq.list_services will return a list of currently available services. The list can optionally be filtered, by service domain and/or service type.

Publishing and Subscribing to asynchronous data streams

Data can be published along with one or more headers which can later be used to filter the received messages:

amq.publish_data(items={'msg': 'Checkout committed', 'order_total': 55, 'products':['book']},
                 headers={'msg_type': 'CUSTOMER_TRANSACTIONS', 'currency': 'USD'})
amq.subscribe_data(subscriber_name='accountant',
                   headers={'msg_type': 'CUSTOMER_TRANSACTIONS'},
                   callback=process_data)

Using logs

TBD