Storage failures do not constitute edge cases. They are inevitable. It loses connections, operators, and networks. This is implicitly assumed by most storage systems partially because it is not likely to occur frequently.
@Walrus 🦭/acc is suited to the time it is.
@Walrus 🦭/acc has redundancy, cryptographic proofs and incentive-based availability, which cause it to keep serving data even when nodes are unavailable. There is separation, dispersion, and retrieval of data as long as the network is within the preset limits. It is not feared that it will fail but failure is expected.
The issue does not lie in the number of nodes that will survive but in the fact that the system can demonstrate that data still exists despite churn. The network level is that guarantee enforced by Walrus.
This is the way the decentralized applications can be used over time. Not because they wish their nodes to remain online, but by designing storage that can withstand when they fail.
