Kademlia File Storage

Kademlia File Storage is a project created for the Distributed Systems course in Computer Science Engineering at University of Padua. It uses Kademlia DHT to distribute file to other peers for storage purpose. Our goal is to achieve the most permanent file storage solution on a p2p network. We will not maintain this project once it's completed (with regards to our course objectives).

How does it work?

When instantiating a node, a FIND_NODE RPC is performed on a list of a handful nodes (which is serialized in the nodes file) to know at most K nodes which are the closest to our node. This file is serialized by the writeFixedNodes method in Kademlia.java, if you need to change those nodes for some implementation/change/whatever purposes, please check that. From then, all other RPCs can be called to interact with the network. While a node is instantiated, it will receive files from other peers which are performing a STORE RPC on the network. These files are frequently refreshed to quickly react to changes occuring in the network, ensuring file persistance for the longest time possible.

How to use it

As you can see in Main.java you can simply join a network by creating an instance of the Kademlia as follows:

Kademlia myNode = new Kademlia();

You can then start using Kademlia 4 basic RPCs (FIND_NODE, PING, FIND_VALUE and STORE) and an extra RPC (DELETE), added by us, to interact with the p2p network and store files. That's it!

If you wish to test it with our provided implementation, just run the start.sh file found in the main directory. This will build all Java classes needed and will run Main.java.

Settings

For testing purposes, we added some settings inside the settings.properties file. These are:

• port: the port address which will be used (MUST be forwarded if behind a NAT).
• fileRefreshThreadSleep: amount of time (in milliseconds) between one file refresh sequence and another.
• fileRefreshWait: amount of time (in milliseconds) to wait before a file can be refreshed.
• bucketRefreshWait: amount of time (in milliseconds) between one bucket refresh and another.
• socketTimeout: self-explainatory (in milliseconds).

Based on Petar Maymounkov and David Mazières paper. Clarifications were found in a presentation made by Tristan Slominski (@tristanls)