There has been a lot of debate about whether Layer 2 or any scaling solutions on Solana won't be of much value. Let’s dive right in.more specifically after a tweet by Mert, It was Sunday midnight, and had nothing else to do so thought why not

Do we need Layer 2s on Solana?

The biggest question when thinking of Layer 2 is whether the base layer is congested or unable to execute all the transactions needed with thousands of transactions stuck in the mempool waiting for their turn.

If we were set out to answer the upper question, no Solana mainnet is functioning pretty well and still hasn't hit its top TPS ever, I know I know what you thinking but doesn't Solana have vote txns and stuff like that

well even after a deduplication Solana as of now should be able to handle upwards of 100k transactions per second but in the real world, it usually hovers around 1k so clearly there is not that high demand for the Solana blockspace right now.Even if the rollups were implemented having solana as their da or execution layer It wouldn't even get close to the data throughput speed of the mainnet even if the rollups themselves are using the Solana virtual machine(fork of Berkeley packet filter).

Credits: Logan Jastremsk

Solana forked rollups on other chains?

That is a good idea to have the fast execution of solana on another chain but here again, we run with da problems, if your dapp is consuming that much compute it would need to dump the data produced somewhere, and right no data availability solution comes close not even thinking about Ethereum.

What about zk?

Sadly adding zk to everything doesn't fix it, Zero knowledge proofs the heart of zk rollups are super compute expensively and take time to be produced that why most of the zk rollups on Ethereum wait hours to batch transactions and post the proof of computation on mainnet.Even if we had parallelized zkp production it would take a long time before it gets to the speed of Solana but one of the pros of that is it would enable a lot faster syncing of Solana nodes that take days as of now.At one point gas fee was higher on zksync and Polygon Hermez because it takes the same amount of compute to make a zk proof and divided amongst the users, if there are few users they would have to contribute to the production of proof

But there is one feature that makes rollups attractive, they provide another level of developer flexibility that is not possible with a single-state machine. Right now every dapp has different requirements and they are made to submit improvement proposals that take ages if accepted and implemented ever but with rollups, devs can just ensure the data availability and conformations of the transactions on some other chain.

The dapp developers might want to have a different fee structure or hope to capture the mev by running their own sequencers for their dapp unlocking another source of revenue in a bear market which would not be possible with mainnet.

PS: Wrote this post at midnight so sorry for any mistakes or typos, if you wanna help feel free to reach out