#Trigger Autoscale with LoadBalancer

##Overall architecture: Use autoscale(otter) api and load balancer api to know when to scale. API calls mediated through pyrax.

##Setup:

1. pip install the requirements: pip install -r requirements.txt
2. open up settings.py and fill in the values that pertain to you
   1. You need to also create a rackspace credential file The "Rackspace Public Cloud" one. Follow the example.
   2. You need to also create a .pyrax.cfg file Instructions. You can look at the example below the graph
3. To run python script: python loadscale.py

If import errors arise. Add this repository to your PYTHONPATH. You can add it to your .bash_profile or .bashrc file. Example: export PYTHONPATH=$HOME/loadscale:$PYTHONPATH

##Testing

1. Go to main directory and type in "nosetests"
2. To get coverage: nosetests --with-coverage --cover-package=loadscale

##Load Testing

(Warning! This will make it work directly with your Rackspace account, incurring costs!): Configure the locustfile.py or just the barebones I provided to run Locust.

1. To run locust -f locust/locustfile.py
2. Then go to 127.0.0.1:8089 in your browser
3. Type in whatever you want for number of users and onboarding speed

##Current Assumptions

1. only one loadbalancer
2. all the same type of nodes in scaling group (which it most likely should be already)
3. load balancer algorithm lends itself to be such that it balances itself out across servers.

##Next Steps:

1. automatically detect optimal connections per server based on flavor (perhaps even through Machine Learning? Or creating some sort of apdex metric). For now, you have to set those values in settings.py
2. Have a better user interface where you can set up account and options.
3. Allow the user to provide a time to wait or have the script run once and kill itself. This way user can control the times they want the script to run rather than having it be an independent process.
4. Based on user requests, this might become more of a streamlined service application that returns values needed by the user. If, for example, they want something that combines both load balancer and autoscale data then they can use this script.