Problem 1: Lack of authorized access mechanisms
Based on the blockchain incentive system, AuroraFS implements allowed data access: nodes can only read the related data from the network when the data owner has granted permission. AuroraFS solution is to encrypt data prior to depositing it in distributed storage, then transfer keys to each other through another channel, and finally use the keys to decrypt the ciphertext read from the storage system. AuroraFS is the world’s first solution for industrial-grade distributed data storage demands, thanks to its authorized access capability to data. Data will be read and transmitted through the next nearby node in the logical distance in existing peer-to-peer file systems such as BitTorrent and IPFS after Distributed Hash Table conducts data addressing. It is impossible to collect real-time data since the optimal nearby node is not chosen based on the network state.
BitTorrent is a peer-to-peer protocol, which implies that computers in a BitTorrent “swarm” (a group of computers downloading and uploading the same torrent) may exchange data without the use of a central server.
Problem 2: Poor performance design
AuroraFS reads data using real-time bandwidth calculation, nearest relay, and multi-clip parallel techniques to improve file reading efficiency, allowing it to serve real-time VOD and live-related data applications. AuroraFS is the basic platform for large data because of its high access performance and ability to grant approved access.
IPFS is a file system that was inspired by the BitTorrent network. Because the BitTorrent network is designed to download data, real-time file reading performance is rarely considered. As a result, the real-time file reading performance of BitTorrents/IPFS has not been optimized.
There is no opportunity to inspect the file before downloading it. You’ll be able to view what’s in the file, but you won’t be able to tell how good it is. The only method to undertake a quality check is to rely on other people’s internet feedback. It is only ideal for applications that do not demand extremely real-time reading, such as file downloads, and cannot be used in situations when exceptional real-time performance is required, such as streaming media, huge data storage, and analysis.
Problem 3: Lack of proper incentive system
AuroraFS introduces blockchain-based storage and traffic incentive algorithms. Participants would be encouraged to provide storage services through the storage incentive mechanism and to provide bandwidth services for the system through the traffic incentive mechanism.
Aiming at the exploitation behavior of nodes in BitTorrent system, a flow model demonstrates the influence of exploitation behavior on system performance, and the “exploitation tolerance threshold” is obtained through this model. When the proportion of exploitation nodes exceeds this threshold, the system may “die”. To avoid system “death “, an indirect incentive mechanism based on encryption is proposed. In this mechanism, nodes must go through a period of seed donation before they can obtain the key and restore the downloaded files. Meanwhile, the mechanism can reward nodes for downloading behaviors according to their seed donation contributions. The experimental results show that the nodes in BitTorrent system will not get any reward for their seed offering behavior, which cannot achieve the effect of suppressing exploitation behavior. After introducing the proposed mechanism, the seed donor node is two to three times more efficient than the original BitTorrent system.
There is no reward mechanism in place. The IPFS incentive system uses the basic BitSwap protocol, which merely calculates the node’s download/upload data and does not use a traffic accounting method. Because there are no traffic incentives, the node has no reason to adopt high bandwidth and make the problem worse.
Problem 4: Lack of fault detection and automatic data recovery functions
AuroraFS presents an “on-demand diffusion” data dispersion strategy, in which stored data is automatically dispersed to neighboring nodes on demand to simplify transmission when it is read frequently and disposed of among adjacent nodes when it is not read frequently.
As for BitTorrent a specialized distributed file system’s error detection and automatic data recovery functions are missing. It forces the application layer to do these essential activities on its own, increasing the burden of application layer development and lowering industrial applications’ desire to use the IPFS file system.
Gauss Aurora Lab
Gold Coast, Australia, 2021
