Mysten Labs's flagship storage project has been pondering the same challenge: how to balance decentralization, replication overhead, recovery efficiency, and security in storage solutions.
Traditional full replication schemes (like Filecoin, Arweave) are simple in logic, but require over 25 times the resources to ensure sufficient security. Erasure coding techniques later reduced this overhead to around 3 times, and projects like Storj and Sia are using them. However, this introduces new problems—recovery costs skyrocket, and vulnerabilities are especially prone to emerge in asynchronous network environments.
The Red Stuff protocol is considered to break this deadlock. It employs a 2D coding plus asynchronous challenge mechanism. Its core idea is based on linear erasure codes and the Twin-code framework, with deep optimizations for Byzantine fault tolerance. How does it work specifically? It splits the data blob into a matrix structure of (f+1)×(2f+1) (corresponding to n=3f+1 nodes, where f is the maximum number of malicious nodes). The primary dimension extends vertically to n symbols, with each node storing a primary sliver; the secondary dimension extends horizontally to form secondary slivers. This 2D design allows the primary dimension to be read quickly with a threshold of f+1, while the secondary dimension supports asynchronous challenges with a threshold of 2f+1.
The most ingenious part is the self-healing recovery mechanism—nodes that lose data only need to fetch row symbols from f+1 nodes or column symbols from 2f+1 nodes, with each symbol roughly 1/n of the original data size.
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
9 Likes
Reward
9
3
Repost
Share
Comment
0/400
ProtocolRebel
· 23h ago
Wow, this 2D matrix design is really awesome. Finally, someone has pinpointed the pain points of erasure coding.
View OriginalReply0
AirdropHunterKing
· 23h ago
Bro, that two-dimensional matrix thing is basically like "storing and playing Mahjong." I've looked at the threshold combinations of f+1 and 2f+1 for a long time, and it feels a bit like the ultimate solution to save gas fees.
View OriginalReply0
Degentleman
· 23h ago
ngl Red Stuff's 2D matrix gameplay is indeed impressive; finally, someone has exposed the recovery cost issue of erasure coding.
Mysten Labs's flagship storage project has been pondering the same challenge: how to balance decentralization, replication overhead, recovery efficiency, and security in storage solutions.
Traditional full replication schemes (like Filecoin, Arweave) are simple in logic, but require over 25 times the resources to ensure sufficient security. Erasure coding techniques later reduced this overhead to around 3 times, and projects like Storj and Sia are using them. However, this introduces new problems—recovery costs skyrocket, and vulnerabilities are especially prone to emerge in asynchronous network environments.
The Red Stuff protocol is considered to break this deadlock. It employs a 2D coding plus asynchronous challenge mechanism. Its core idea is based on linear erasure codes and the Twin-code framework, with deep optimizations for Byzantine fault tolerance. How does it work specifically? It splits the data blob into a matrix structure of (f+1)×(2f+1) (corresponding to n=3f+1 nodes, where f is the maximum number of malicious nodes). The primary dimension extends vertically to n symbols, with each node storing a primary sliver; the secondary dimension extends horizontally to form secondary slivers. This 2D design allows the primary dimension to be read quickly with a threshold of f+1, while the secondary dimension supports asynchronous challenges with a threshold of 2f+1.
The most ingenious part is the self-healing recovery mechanism—nodes that lose data only need to fetch row symbols from f+1 nodes or column symbols from 2f+1 nodes, with each symbol roughly 1/n of the original data size.