--------
overview
--------
We use flask to implement a simple loadbalancer. 

In the current architecture, the front-end app sends requests to one
production machine running the code from gridspice.webserver 
(see code.google.com). 

The idea is to have the front-end app send requests to a load-balancing 
machine (this code). The load balancer, in turn, distributes requests
among the many machines running the gridspice.webserver code. 

To run, simply type -

$ python2.6 index.py

------------
dependencies 
-----------

Must use python2.6

1) easy_install-2.6 elementtree
2) install flask, a python web framework. 

-----
Files
-----

index.py: handles url routing. This is a thin wrapper around calls to
SimulationManager.py

SimulationManager.py: Defines a class for managing simulations. 
There are 2 main methods:
    - start_simulation, which is called to create a new simulation. It takes 
      the xml file that came with the request, splits it up into files with
      only one model per request, and forwards the request out to a child
      server (i.e. the loda balancing part). 
    - handle_generic_request, which is called to handle any other request. 
      Looks up which server the simulation with the corresponding
      simulationId/modelId is being run, and forwards the request out
We store a mapping of simulations to the server the request is being run on
on the filesystem. We create a directory called simulations, and then in
the simulations directory we create a folder for each simulation id. within
each of these folders, there is a json file named by the model id which
stores the host and the port where the models is actually being run. 


server_config.json: used to specify config options for the load balancer.
Right now, it just hardcodes a list of servers that can act as children.

test.xml: an outline of what an xml file that comes from the front-end would
look like