ACDS in the Walrus protocol ensures that core storage remains safe and reliable even in asynchronous networks and in the presence of Byzantine (malicious or faulty) nodes. In Walrus, data is divided into multiple parts using erasure coding, where each honest storage node stores only its assigned fragment instead of keeping a full copy of the data.
This design significantly reduces storage costs while still guaranteeing that the data remains recoverable. ACDS provides a strong consistency guarantee: two honest readers will never receive different results—either both successfully retrieve the complete data, or both fail to do so. If the writer is honest and the data is successfully written, then any honest reader can later read that data.
Walrus uses ACDS together with epoch-based committees, where storage nodes rotate over time. This rotation improves scalability and strengthens security across the network. Overall, this approach makes Walrus a practical, low-cost, and Byzantine-resilient decentralized storage solution.
