I am reading the NL blogposts again and regarding the one million transactions per second (TPS) on Tezos claim, one questions comes to my mind and I think I am missing some pieces here.

As far as I understand it is not possible to use Tezos in a rollup. If I want to I need to wrap it as rollups cannot have Tezos balances but only tickets. So how are transaction/storage fees handled on a rollup? If I run for example 1000 rollups that run 1000 TPS - the rollups are sending TXN´s between each others.

How are the fees paid as rollups only can hold tickets?

1 Answer 1


In a rollup, by default, you have neither tez account nor tickets.

A rollup is a child chain that can process what one will code in its kernel. L1/L2 communication via inbox/outbox is only bytes.

The kernel can be design for one specific app (app chain model) or can be a smart contract VM: there is currently teams working on EVM Rollup & Michelson Rollup

Tickets are a way to manage assets promoted because it represents a right on an assets with strong invariants provided by Tezos. So developping a kernel with a ticket ledger avoid some complications of classical two-way bridges. One could even develop an account model in its kernel. This is not mandatory and a rollup developer decides its asset model. Rollups kernels are unopiniated.

For writing in inbox or outbox, one pays for XTZ fees in L1. There is nothing implemented for L2 fees, the rollup creator decides and develops its fees model, if any.

It is important to understand that Smart Rollup (aka SORU) is a solution to design permissionless & censorship resistant rollup. It is not a specific implementation of a rollup (unlike TORU or competitor rollups)

If I run for example 1000 rollups that run 1000 TPS - the rollups are sending TXN´s between each others.

There is currently no L2-L2 communication. 1MTPS assumes there is no dependencies between the transactions on different rollups (parallel computing) or that they are aggregated in L1 (map/reduce computing)

  • Interesting, my picture how I imagined smart rollups was a bit different. But due to some misunderstanding on my side. But regarding the L2-L2 communication is this just temporarily? So with the assumption that there is no dependency between the transactions on different rollups - the rollups are sending TXNs but not between each others - so every rollup is a kind of "closed" system on its own?
    – David
    Commented Mar 20, 2023 at 23:27
  • 2
    It's not closed, you can already do L2->L1->L2 but it means you have the 14 days latency. The first solution is a social answer: a) one can build a "fast withdraw system" for FT against some fees; b) rollup will build its trust: a dishonnest rollup will have bad social ranking and will not be used. You could build Oracle systems with rollup you trust; c) There is possible technical answers, but it will take time cause it is not easy and it lead to other topic like normalizing messages accross rollups
    – Thomas
    Commented Mar 21, 2023 at 15:18
  • I can understand that L2 -L2 communication is not possible via L1 chain (inbox/outbox messages), but a system can be designed where rollups can communicate with each other without L1 being in the picture? Is this possible? Commented May 4, 2023 at 11:48
  • You can implement atomic swap, that's basically what layerOne does. One need to provide liquidity for it and maybe insurance system. But it cames also with risks: if ROLLUP A is accused and refuted, then ROLLUP B will
    – Thomas
    Commented May 4, 2023 at 14:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.