However, loss of data usually occurs not because the network is down but because of the design of redundancy. Walrus focuses on this by controlling how the replicas of data are distributed within the network, unlike current implementations where it is up to the individual storage companies. When storing data, @Walrus 🦭/acc makes sure it distributes pieces of data across separate nodes.

Instead, it ensures each replica is located within different failure domains rather than multiple replicas within a similar environment. Therefore, hardware failures, geographical issues, or node separation will not impact multiple replicas at the same time. Consequently, data will still be available whenever some domains within the network experience instability.

$WAL is also constantly checking the status of the replicas, contrary to the redundancy of replication that is assumed to exist for all eternity. When a node goes down or no longer satisfies the availability need, the network is set to perform a re-replication automatically, and additional storage providers are assigned to raise the redundancy level to the needed amount.

The incentives for WAL are deeply linked to such a system. Providers earn their rewards as long as they continually support the replicas allocated to them and reach the availability thresholds. The provider’s reward will decrease if the replica is no longer available or if the replica’s performance is degraded, thus encouraging the provider to offer high availability.

From the perspective of the user, this means that the stored data is not dependent on static assumptions regarding the reliability of the nodes. Redundancy is enforced by the protocol itself. Instead of degrading with the addition of nodes or subtraction of nodes, the availability of the data improves with time.

Walrus achieves this by using a combination of replica placement control, continuous observation, and incentives using WAL. This makes the redundancy in the system a living system instead of a fixed setup. This increases the robustness of the network as it scales. #walrus