MaidSafe

The MaidSafe network is an open-source, decentralised, autonomous network for data storage and retrieval by end user applications. All data is stored on the network within vaults, which are member nodes of a self-managed network resembling a distributed hash table (DHT). We explore the design of the vault network, including the self-managing nature of both vaults and data, and attack vectors worth consideration and further research.

Today all known mechanisms that grant access to distributed or shared services and resources require central authoritative control in some form,raising issues in regard to security,trust and privacy. This paper presents a system of authentication that not only abolishes the requirements for any centrally stored user credential records,it also negates the necessity for any server based systems as a login entity for users to connect with prior to gaining access to a system.

Today’s Distributed Hash Tables (DHT’s) and other overlay networks are based on operating without hindrance of real world issues regarding connectivity between nodes. This is not a problem when operating in a private or controlled environment,but in the transition to peer to peer or fully distributed networks,it becomes a major headache. This paper introduces a pure p2p solution to Network Address Translation (NAT) traversal,which is probably the main problem facing public p2p networks.

An effective distributed network requires an addressing mechanism and distribution of data that is designed to overcome churn (the unmanageable outage,node failure or unforeseen communication fault). There are several implementations of such a system,but this paper proposes a system that not only provides an efficient churn-resistant DHT,but also has fast awareness of infrastructure changes. This allows the implementation of a locking system,thereby enabling amendment of data from multiple sources.

Autonomous networks are self- healing,self-managing and most importantly independent of human interference. Such networks will be able to be developed in a way that avoids wasting effort on maintaining even simple mechanisms such as storage,scalability and data retention. Sys- tems like these will quickly extend to providing a method of highly scalable platforms that can accommodate real time transactional logic. A working example of an autonomous network is outlined in this paper.

Distributed file systems require servers or control nodes. Access to a file system is a security issue that can apparently only be controlled by some kind of authority,and this is always a potential point of failure. These file systems also require an indexing mechanism. This paper presents a distributed file system without centralised control or indexing. This file system also utilises a distributed locking mechanism to ensure data integrity in the case of multi-user access to any file.