High-level architecture

Data Availability

The DA service is used to store transaction data and make it available to the network. The service is accessed via DA light nodes.

Celestia is used as the Data Availability (DA) layer.

Sequencing

In order to protect the DA layer from spam, only certain full nodes receive authorization to sequence transactions in ordered transaction-batches. These full nodes provide basic sequencing for transactions ingressed from the users. The full node then sends the transaction-batch to the DA service using the DA light node.

The DA service sequences transaction batches (which have the form of data blobs). This results in proto-blocks, and where each proto-block may contain several transaction-batches. The proto-block is the final ordering step of transactions. Next stop: execution.

Execution

The full node receives proto-blocks via the DA light node and executes the transactions in the proto-blocks. The full node then creates L2 blocks from the proto-blocks, which contain additional information such as the new state root after applying the transactions.

Settlement

The network uses the Fast Finality Settlement Module for transaction finality.

Validator nodes check against the postconfirmed state root on L1 to increase their security and reduce the risk of providing wrong state updates to users. Initially - with the training wheels in place - a specialized node, called Core-Validator, is responsible for updating the state on L1.

As the training wheels are removed the network will move to a decentralized model where validators will be able to attest to the correctness of state transitions, and a supermajority of validators will be required to advance the state roots on L1.

Fees

Transaction fees are paid in the token of the L2 network. They are used to pay for the data availability, sequencing, execution and settlement services:

fee = data_availability_fee + sequencing_fee + execution_fee + settlement_fee

For more information on fees, see this MIP.