Walrus Write Flow: Every Node Gets Its Pair + Signed Proof of Storage
Write operations in Walrus follow an elegant protocol that ensures accountability while maintaining efficiency. When a client writes a blob, it doesn't broadcast to all validators.
Instead, fragments are delivered to specific validators determined by 2D grid structure.
The write flow is deliberate: each validator receives its designated sliver—the specific fragment it's responsible for storing. For primary slivers, this is direct blob data. For secondary slivers, this is derived redundancy. The protocol computes which validator gets which fragment from grid position and blob ID.
Critically, each validator that receives a fragment returns a cryptographically signed proof of storage. This proof, recorded on-chain via Sui, becomes an immutable record that the validator accepted responsibility for that specific blob and fragment. The signature proves the validator's explicit commitment.
This proof-of-storage model is more powerful than traditional approaches. A client has cryptographic evidence that specific validators hold specific data. If a validator later claims data is lost or unavailable, the signature is proof of dishonesty. Disputes are resolved on-chain with mathematical certainty.
The signed proof also enables economic incentives. Validators that consistently provide proofs earn reputation and fees. Validators that refuse or delay proofs face economic consequences. The on-chain record is transparent and enforceable.
This write flow combines efficiency—no broadcasting—with accountability—every fragment has a verifiable signature. The result is reliable storage where dishonesty is impossible to hide.
