In the ZK-rollup ecosystem, the infrastructure includes nodes, sequencers, provers, and more. In Ethereum Layer 1, there are validators but no provers. Flashbots can connect MEV searchers with validators.
ZK-rollup or Layer 2 (L2) has some differences. It has provers, and each block must be proved and have a proof.
Here is a landscape of the ecosystem based on our understanding, using the Taiko project as an example. The Taiko project defines itself as a based rollup. The user experience is no different from L1, including using ETH as gas fee.
In L2, MEV follows a similar flow. MEV searchers can visit the mempool of L2 and extract value. However, for the MEV capturer of L2, it's not enough to just build a transaction bundle of L2. The transaction proposing an L2 block needs to be guaranteed to be included in an L1 block, or the value of L2 MEV cannot be captured. Therefore, an L2 MEV searcher should also be an L1 MEV searcher. (Reference 2)
Another difference is that each block needs to be proved. In Taiko's design, before a block is proposed, it should be bound with a prover. Here, ZKPool plays the role of a prover pool. ZKPool can provide the price of a block, and the sequencer can choose and then the contract in L1 can confirm the assigned prover. Finally, ZKPool captures the event and starts scheduling a prover to do the work.
To summarize, L2 has an infrastructure ecosystem consisting of MEV searchers, ZKPool, and provers. Within this framework, MEV captures the value and may become the sequencers of L2. Provers contribute proof and get rewards from sequencers. ZKPool collects prover resources and selects the most efficient ones with a favorable performance-price ratio.