0G Removing DAS
Introduction
0G uses a quorum-based model for data availability confirmation.
This provides 0G Storage, which is responsible for storing data on-chain, with the ability to verify data availability in a timely and secure manner. Our quorum-based architecture is unique in the sense that it uses a VRF to randomize the quorum, emphasizes a reward-based design, and uses GPUs to accelerate the erasure coding process.
Below we’ll expand on this decision and discuss its implications.
A Quorum-Based Model For Superior Performance and Security
Data Availability Sampling (DAS) is a common approach used by DA Layers, entailing the sampling of various portions of data to verify that all data exists. However, it comes with limitations from an efficiency perspective.
For example, Celestia’s 12-second block time is due to their need to optimize light client bandwidth for DAS purposes. 0G uses erasure coding which entails data replication, which would require cross-checking for utmost security. Without cross-checking, there could be time for a malicious party to fool a DA node or DA client.
0G instead uses a quorum-based model with an “honest majority assumption”, which requires nodes to agree that data is available. At the same time, at least one additional entity holds a copy of the data, meaning that it can be verified if needed (similar to an optimistic rollup approach).
There are 3 unique aspects to this approach:
- Verifiable Random Functions (VRFs): This takes a secret key and some input, and produces a random output. 0G uses VRFs to construct the quorum of nodes, mitigating any potential for collusion.
- Reward-Based Design: Rather than using slashing to punish, 0G focuses on rewarding participants.
- GPU Acceleration: We use specialized GPUs to accelerate the erasure coding process, increasing the data throughput rate for each DA client.
We believe that this approach will maximize both security and performance, which is critical given 0G’s role as both an on-chain data storage and data availability solution.
To learn more and join our newsletter, visit https://0g.ai/ or follow us on Twitter.
