[Feature] EIP-4844 Blob support

[madumas]

Description

Proto-danksharding is slated for deployment on the Ethereum mainnet in March, following its successful implementation on test networks. The primary focus of EIP-4844 has been on Layer 2 (L2) solutions, which will utilize the new Type 3 transactions for data storage, rather than depending on the (comparatively) expensive calldata. However, it's anticipated that various other applications will also adopt this innovation. There's already discussion among DeFi dApps about exploring this capability.

We don't know how it will be used, but graph-node should support blobs and allow subgraphs to parse and decode the data to make it usable by developpers. This feature would allow developers to access the information without needing to resort to potentially costly RPC calls. The Graph might also offer an effective solution to the data retention challenges posed by EIP-4844.

Several technical questions need addressing:

• Determinism: By default, consensus layer (CL) nodes will only store blob data for a very brief period. This implies that subgraphs indexed at different times will capture varying data sets. The network lacks support for synchronizing this data, but indexers might have the option to retrieve historical data by other offline means, though this is yet to be confirmed.
• Consensus Layer (beacon chain/eth2) Support: At present, graph-node does not support RPC calls to CL nodes.
• Standardization of Blob Content Format: The format for blob content has not been standardized.

Are you aware of any blockers that must be resolved before implementing this feature? If so, which? Link to any relevant GitHub issues.

No response

Some information to help us out

• [ ] Tick this box if you plan on implementing this feature yourself.
• [X] I have searched the issue tracker to make sure this issue is not a duplicate.

[azf20]

Hey @madumas many thanks for this. You can read more about current ecosystem support for EIP-4844 here. The Blobs subgraph leverages a new consensus-layer firehose (where indexers run "non-pruned" Consensus clients as necessary), and a substreams-powered subgraph. Are you seeing specific use-cases emerging, beyond adoption of this new transaction type by rollups?