Increasing Rollup TPS to 100k and more.

There are currently two bottlenecks:
1) the data availability (DA), and
2) the execution.

Data Availability (DA)

With the introduction of blobs (EIP-4844), the cost of posting transaction data has been reduced by 17x.

Before EIP-4844, posting 128kb of Ethereum call data cost $76, and this changes with network demand.

With Blobs: Posting 128kb costs only $4, and this will increase as more rollups and app-chains start using blobs.

What can Rollups do to reduce DA cost and pack more data?

Compress the transaction data before posting: According to Vitalik’s post, a transaction takes up around 180 bytes, but this can be reduced to 25 bytes using compression techniques.

This will allow layer2s to reduce the gas fee by 7x.

What can Ethereum do to reduce data posting cost?

Increase the number of blobs per block. This update can be done relatively easily by a simple parameter change.

EIP-4844 (aka “blobs”) does not give us data availability sampling. But it does set up the basic scaffolding in such a way that from here on, data availability sampling can be introduced and blob count can be increased behind the scenes, all without any involvement from users or applications. In fact, the only “hard fork” required is a simple parameter change. — Vitalik

Today, the bottleneck is Execution because the demand for Blobs hasn’t picked up yet.

What can Rollups do to improve Execution layer Throughput?

a) Increase the read and write speeds:
As the state grows larger, lookup times for any given piece of state become larger, which in turn increases the transaction execution time.
Rollups should innovate new data structures that make state access (reading values) and state updates (updating values) cheaper and faster to make state growth more tenable and increase transactions per second (TPS).

Example:

Erigon/reth offer some improvements to database efficiency, and some rollups have clients based on these clients (in particular op-reth). Erigon/reth use a flattened data structure, which reduces lookup costs somewhat for reads; however, there isn’t support for asynchronous reads or multithreading. Also, the merkle root still needs to be recomputed after each block, and this is quite slow.

b) Parallelization:
Rollups could implement something like EIP-648. This is an important engineering challenge for rollups.

c) Improving hardware efficiency:
Some teams don’t focus on this, arguing that if the node requirements are high, it would centralize the system. However, even if the nodes are centralized, node operators still can’t submit wrong transactions. so, team should work on improve the efficiency from hardware.

Finally, as new app-chains are launched, the demand for Blobs increases, and they will become the bottleneck if Ethereum doesn’t increase the number of blobs per block.

TLDR: Layer2 team need to work on data compression, Increase the read and write speeds, Parallelization and Improving hardware efficiency to increase TPS and reduce gas fee.