Fraxtal DA implementation consists of a modified op-batcher and op-node executable that uses our custom DA implementation as the storage backend of the chain data.

Our custom implementation along with the other node changes can be found in this (opens in a new tab) repo.

The major changes are in the op-batcher which is the component that’s responsible for storing the chain data and posting the reference on the L1 to be picked up by op-node.

Compared to the default OP implementation where data is stored in EIP-4844 blobs, ours sends the data to a da-server which is responsible of uploading the data on multiple backends and returning its IFPS hash. The IPFS hash is then posted on the L1 so it can be picked up later by op-node.

We are currently hosting the data blobs on 4 different storage backends with a total of 12 copies.

The reference posted on the L1 is then read by our custom op-node implementation that fetches the actual data from a DA endpoint (we provide a public one but anyone can build one that relies only on IPFS).

Future plans

For the future we’re planning to expand our supported DA backends/implementation. Some of the options we’re reviewing are:

• EigenDA
• Celestia
• Near DA
• EIP-4844 Blobs
• OP plasma integration