SETUP:
======

1. Ubuntu/debian will need the following package installed. You can install them using apt-get:

	Python 2.6
	python-webpy (also: http://webpy.org/install) version 0.34
	python-nose
	python-paste
	python-mysqldb

2. Then get mimerender and mimeparser packages from:
	http://mimerender.googlecode.com/files/mimerender-0.2.2.tar.gz
	http://mimeparse.googlecode.com/files/mimeparse-0.1.3.tar.gz
	
3. extract them and install using:

	$ python setup.py install

4. Run the provided database definition script in project home directory with your mysql server:
	
	$ mysql database_definition.sql 
	
	give credentials where necessary.

5. Go Through configuration file (configuation.cfg) and tweak the user name, passwords, other credentials
	as you deem fit. The store can be configured to run on multiple mysql backends on different hosts. 
	 	
6. Make controller.py and sync.py executeable by running command:
	$ chmod +x controller.py sync.py

7. Pass all tests using:
	$ WEBPY_ENV=test nosetests
	
8. Add sync.py to your crontab to run every few minutes
	
9. Now, you can run the web server by just running controller.py file from the commandline using:
	$ ./controller.py


STRUCTURE:
==========

-	This data store is divided between data layer and control layer (i.e Model and Controller part of MVC.

	controller.py contains all the control layer and is responsible for fielding all the request and
	returning well formed HTTP validated responses. 
	
	datastore.py has the data layer code. It is responsible for data manipulation. Storing and retrieving 
	from the back end. 
	
	sync.py is syncronization script which handles the lazy synchronization of key value pair across 
	all the nodes. It is supposed to run lazily on a frequency of choosing.

	
TESTING:
========

	Running the followig command on the base folder would execute all the test:
	$ WEBPY_ENV=test nosetests
	
	
HOW TO USE:
===========

	The test folder contains test for two types of access. They are the best documentation on how to use the service.
	Tests in the files GET and SET key value pairs. As well as UPDATE, DELETE them. You can also list all the
	keys in the data store. All the successs and error codes are HTTP compliant.


NOTES:
======

-	The current installation would be for testing purposes. However, you can easily configure it to use
	some serious webserver. The instructions to use webpy in production with LightTPD and Apache can be
	found at:
	
	http://webpy.org/install#prod

-	The service is configured on two back end nodes. It is very easy to scale beyond that. Just change the
	value of partitions in configuration file to desired number of nodes. Then add section for configuration
	for particular node. The section should be named following conventions of python configuration file. For
	example: partition-3, partition-4 and so on.
	
-	The front end is http webserver can also be scaled independant of the back end. However, sync.py should 
	only have one instance to avoid race conditions.
	
-	TEST mode requires all the database nodes to have 'test_' in the start of their name. So it is 
	imperative that we have a database with name 'test_' available on the same configuration. This is important to
	keep in mind if we want to change mysql node configuration.