#blueshed.micro The problem is how to use Tornado with SqlAlchemy and scale. The answer is python. Since python 3.4 there is a ProcessPoolExecutor class which manages a pool of subprocess to which you can submit functions. All your blocking requests are submitted to the pool, or you can have multiple pools for fast and slow functions.
You can have multiple processes, schemas, databases, tornados... To have websockets that broadcast state changes to their clients you'll need a way of telling each tornado of the change. So a nice async service is Pika. Tell Pika that each tornado should broadcast your change.
The micro websocket and handler pass a broadcast queue to the subprocess in a context, and handle cookies and authentication that way. The context returns unless an exception is raised. By using annotations you can specify if your function requires a context.
Tornado is brilliant at server side templates and especially websockets. SqlAlchemy is brilliant at complex, optimized sql driven by a domain model. Python is brilliant because it enables both to be possible and to work well together. pytest-tornado is another must have. It allows you to test tornado application, even websockets, without coding anything more complex that the hello world example on the tornadoweb.org index page.
Still to come, memcache. MySQL is brilliant too as it provides a memcache interface to innodb tables. It means SqlAlchemy can manage your cache updates within its transactions or schedule a persistent task for it to be updated.
references:
Below are some sample projects that explore these ideas. They also include a NoSQL example.
heroku bound scalable chat
To run the chat example
python -m spddo.chat.server
Now with a micro service framework.
To run the example micro:
python -m spddo.micro.server
There is also a demo of how to work with Motor and Mongodb using fullCalendar's Scheduler.
To install you'll need brew:
/usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
and to install mongodb:
brew install mongodb
then you'll need to install the tornado mongo client:
pip install motor
then to run:
python -m spddo.mongo.server