EIP-4844 introduces a new method for reducing Ethereum's transaction costs by utilizing blobs, which provide a temporary and cheaper way to store rollup data compared to traditional calldata.

By minimizing the use of Ethereum's block space and only posting essential data, rollup transactions become significantly cheaper as they are executed and stored off-chain.

These blobs are attached to transactions but are not permanently stored on the Ethereum blockchain, which helps in reducing long-term storage costs and freeing up valuable blockspace.

The primary goal of rollup chains is to reduce congestion on Ethereum, lower gas fees, and maintain security and decentralization by relying on Ethereum for finality.

When a rollup chain submits transactions, it sends a blob-carrying transaction that includes a proof and the blob to Ethereum, allowing validators to verify the transaction without needing to re-execute all the operations.

Each blob can be as large as 1 MB, and an Ethereum block can accommodate approximately 64 blobs, depending on the overall size and gas limits.

Once the blob data expires, it is no longer accessible on Ethereum; however, its validity can still be confirmed through the associated proof and hash reference.

Validators can distinguish between regular transactions and blob transactions based on their structure, with blob transactions incorporating proof and temporary data.

Rollups like Arbitrum and Optimism serve as Layer 2 solutions that process transactions off the main Ethereum chain, enhancing speed and reducing costs.

After a designated period, the blob data is discarded, but a hash reference remains in the Ethereum block to facilitate transaction verification.

Traditional Ethereum transactions continue to use established methods that require permanent storage for use cases needing immutable records, while blob transactions focus on scalable, temporary solutions.

Prior to the implementation of EIP-4844, rollups stored their transaction data permanently on-chain, which led to increased costs and limited scalability.