Properties of HotShot

HotShot’s design intentionally aims to give L2s fast finality guarantees, while being able to scale to a large number of participating nodes. However, the participating nodes do not execute transactions; hence, individual nodes only need assurance of data availability to vote in consensus, not to have full access to the data. This alleviates high hardware requirements for participation, without sacrificing throughput.

HotShot is based on the consensus techniques used within HotStuff and HotStuff-2.

For more details on HotShot and Tiramisu DA, see Designing Espresso: Combining HotShot Consensus with Tiramisu DA.

Key Properties of HotShot

Separating data availability (DA) and execution from consensus

The HotShot implementation is purpose-built for providing fast finality to a large number of generic L2s. In particular, it does not perform execution, and the data availability requirement (i.e., ensuring that the system has access to data) is implemented by a separate DA layer called Tiramisu. This enables HotShot to process more data than typical state machine replication protocols. Such modularity also allows the use of various appropriate sub-protocols as needed.

Scalability

HotShot relies on all-to-leader and leader-to-all communication, thus reducing the consensus communication complexity to linear in the number of nodes. Since HotShot does not require every node to get a full copy of transaction data, low consensus communication is especially important. HotShot combines this optimistically with a content delivery network (CDN) to efficiently route data and perform computation. This reduces the leader bottleneck and supports a system with a heterogeneous set of nodes, without sacrificing safety and liveness guarantees. These improvements will help HotShot to scale to thousands of nodes, such that it can be run by a large number of Ethereum validators through restaking.

Responsiveness

HotShot is optimistically responsive and thus, under favorable conditions, commits new blocks as fast as the network allows. This ensures that the protocol’s performance is directly related to the state of the network—under optimistic conditions, the protocol can have low latency and consequently high throughput, too. In HotShot, using a CDN at the network layer synergizes with the optimistic responsiveness property to provide even better performance.