Vitalik: Fusaka is developing the core feature PeerDAS, aiming to enable real-time blockchain without the need to download complete data.

Jinse Finance reported that Ethereum founder Vitalik Buterin posted on X, stating that for Fusaka, security is of utmost importance. Its core feature, PeerDAS, is attempting something unprecedented: creating a real-time blockchain that does not require any single node to download all the data. The way PeerDAS works is that each node only requests a small number of "chunks" of data, probabilistically verifying that more than 50% of the chunks are available. If more than 50% of the chunks are available, then in theory, nodes can download these chunks and use erasure coding to recover the remaining data. In the first version, the complete data of the block still needs to exist in one place, in two scenarios: (i) initial broadcasting: when the data is first published; (ii) reconstruction: when the publisher has released between 50% and 100% of the data chunks. However, these roles are all trustless: we only need one honest participant to perform these tasks, and even if there are 100 dishonest participants, the protocol will bypass them. Moreover, different nodes can perform this task for different blocks. In the future, cell-level messaging and distributed block building will allow these two functions to also be decentralized. All of this is new technology, and it is wise for core developers to remain extremely cautious in testing, even though they have been working on this for many years. This is also why the initial number of blobs will be increased conservatively, and then become more aggressive over time. But this is key to L2 scaling (and ultimately also to L1 scaling, once the L1 gas limit is high enough that we must put L1 execution data into blobs).