snakes is a very basic distributed key-value store written in Python. With snakes, you can run multiple nodes that are all able to act as read and write masters. Nodes will attempt to keep each other up-to-date. snakes is merely a prototype and therefore has many limitations. Below you will find a list of snakes basic features and limitations:

• snakes communicates over HTTP, both with clients and among nodes. To interact as a client, you issue simple HTTP requests.
• Keys must be alphanumeric strings
• Values should be strings of limited size (snakes will hold and compare values in-memory). The contents of the values are opaque to snakes, it can be anything (e.g. JSON, XML)
• snakes uses a very simple file storage. It uses one file per key. This keeps snakes simple but will not result in optimal performance as snakes will have to do a lot of seeks. SSD drives are recommended if you for some strange reason wants to use snakes under load.
• snakes does not currently support any authentication. However, snakes is meant to be placed behind a reverse proxy (such as Apache httpd or Nginx) where authentication could be performed.
• snakes has very limited logging
• snakes does not support any form of transactions. However, a future improvement might support optimistic locking with the use of HTTP e-tags.
• snakes does not support partitioning
• Nodes in a snakes cluster will sync updated among themselves, but only as long as nodes are available. During node downtime or network partitions, nodes might get out-of-date. Nodes will not actively try to sync themselves after such an incident. Instead, snakes relies on clients to update values. snakes uses a simple quorum algorithm (last update wins) to determine which value to serve a client. A future improvement might provide the client with all existing values (and update times) to choose among in order to merge a conflict.

snakes is very simple to start. You can run as many nodes as you like. Each must have a unique IP address/port combination and a separate storage directory. To start three nodes on the same server, run:

python3.1 snakes.py http://127.0.0.1:9001 http://127.0.0.1:9002
python3.1 snakes.py -p 9001 -d data2 http://127.0.0.1:9000 http://127.0.0.1:9002
python3.1 snakes.py -p 9002 -d data3 http://127.0.0.1:9000 http://127.0.0.1:9001

snakes requires Python 3.1.

snakes will default to running on port 9000 and use a data directory named "data". To provide a different port, use the -p option. To provide a different data directory, use the -d option. You must also provide the URLs for all other nodes in the cluster. A future update of snakes might instead use mDNS to automatically locate nodes in a cluster.

To communicate with snakes you use simple HTTP requests.

To get the value of a key, issue an HTTP GET request to the server with the key as the path. In the examples below, the key name is "foo". The examples below uses curl, but any HTTP client can be used:

curl http://localhost:9000/foo

This request will return the value as the content of the response. If the key is not known to the cluster, a 404 response code will be returned.

You can also configure how many nodes in the snakes cluster that should be ask to provide a value for the key. You do this using the r query string parameter. By default, r=3. Here's an example:

curl http://localhost:9000/foo?r=2

To store (create or update) a value for a key, issue an HTTP PUT request, for example:

curl -X PUT -d "some value" http://localhost:9000/foo

On a successful update, a 201 response code will be returned.

On updates, snakes will update all nodes in a cluster. Some of these will be done synchronously before returning a response to the client. The remaining will be done on a background thread pool. By default, 3 nodes gets updated synchronously. However, you can tell snakes if you want a different number of nodes updated synchronously by using the "w" query string parameter. For example:

curl -X PUT -d "some value" http://localhost:9000/foo?w=4

That's all there is to it. Now, go play :-)