Why is Rollups + data sharding the only way to achieve one million TPS?

The reason for supporting Rollups + data shards is that the combination of the two is more secure and decentralized . But this is only part of the story. Rollups + data sharding is the only solution to achieve global scale, and the real reason lies in the scalability brought by the combination of the two : this is the only way to achieve millions of TPS (transactions per second) in the long run. Specifically, I will explore zkRollups because Optimistic Rollups have inherent limitations.

So why do you say that? The reasons can be attributed to two points: a) technological sustainability ; b) economic sustainability .

Technology sustainability

A technically sustainable blockchain node must do three things:

  1. Follow the blockchain and synchronize the nodes;
  2. Able to synchronize from the genesis block in a reasonable time.
  3. Avoid losing control of state expansion.

Obviously, for a decentralized network, all the above three points are essential, but they also bring serious bottlenecks. While the Ethereum network guarantees all these three points, it is advancing the edge of possibility, which is obviously not enough: a sharded chain that retains these three points can only increase the scalability to a few. Thousands of TPS is still not high enough .

Centralized solutions and their limitations

But some more centralized blockchain networks may begin to compromise: 1) You don’t need everyone to follow the blockchain, as long as there are a minimum number of validators; 2) You don’t need to start from the genesis area. Blocks start to synchronize, just use snapshots and other shortcuts; 3) State expiry (state expiry) is a good way to deal with state expansion, which will be implemented in most blockchains, before that , Solutions such as regenesis may be helpful. So you can see that these blocks chain network is no longer a decentralized, but this article is not concerned to the center, we are only concerned with scalability (scalability).

For these networks, the first) compromise constitutes a hard limit. RAM (memory), CPU, disk I/O, and bandwidth are potential bottlenecks for each node, and more importantly, keep the minimum number in the network The synchronization of nodes means that there is a hard limit on how far the blockchain can go. In fact, you can see that networks such as Solana and Polygon PoS are already very hard, even though they can only handle a few hundred TPS.

I visited the Solana Beach website, the Solana blockchain explorer, and it says “Solana Beach has trouble keeping up with the Solana blockchain.” The block time shown above is 0.55 seconds, which is less than the target block time of 0.4 seconds. An increase of 43%. According to Solana’s technical documentation, you need at least 128 GB RAM to keep syncing with Solana, but even 256 GB RAM is not enough to start syncing from the genesis block, so you need to use snapshots. This is the 2nd) compromise mentioned above, but we will ignore this aspect for the time being in this article and focus only on scalability. Jameson Lopp (Bitcoin Core developer) tested it on a 32 GB RAM machine, and as expected, the Solana node crashed within an hour. Of course, Solana is a good example, but so are some other blockchains.

zkRollups can well surpass centralized L1s

zkRollups may even have higher requirements than centralized L1s, because the validity proof of zkRollups makes it as safe as the most decentralized L1! In the zkRollup network, you can have only one node active at any given time, but this is still highly secure. Of course, in order to make zkRollups censorship-resistant and resilient, we need multiple sequencers ( Sequencers), but even these sequencer does not need to reach a consensus, they can rotate accordingly. For example, Hermez and Optimism plan to have only one active sequencer at a time, and will rotate between multiple sequencers.

In addition, zkRollups can use all the innovations to make full-node clients as efficient as possible, regardless of whether these innovations are for zkRollups or L1s. zkRollups can be more innovative through state expiration technology, because historical data can be reconstructed directly from L1. In fact, there will be innovations about sharding and historical data access pre-compilation, so that zkRollups can be run directly on data shards! In zkRollups, we also need light and unassisted withdrawals to ensure its safety.

However, even in zkRollups, we will encounter hard limits. Regardless of whether it is 1 TB RAM or 2 TB RAM, there is a limit to how far a node can go. At the same time you also need to consider the infrastructure providers, they need to be able to keep pace with the chain.

So, yes, some may zkRollup scalability L1 much better than most more scalable , but relying on its own can not achieve global scale .

Run multiple zkRollups through data sharding

But we can run multiple zkRollups on Ethereum’s data shards . Once released, they will provide a large amount of data availability, and will continue to expand as needed, and are expected to reach 15 million TPS by 2030. A single zkRollup cannot achieve this crazy throughput, but multiple zkRollups can.

Will each zkRollup shard break composability? For now, it will. But we are seeing a lot of work in this area, such as rapid “bridges” such as Hop, Connex, and Biconomy, and outstanding innovations such as dAMM that allows multiple zkRollups to share liquidity. Many of these innovations will be more difficult or impossible on L1s. I expect continued innovation in this area, enabling multiple zkRollups chains to interoperate seamlessly .

In short, no matter what the most centralized L1s can do, zkRollup can do better, and its TPS is much higher. In addition, we can have multiple zkRollups, which can effectively reach a global scale.

Economic sustainability

This is easy to understand. The network needs to get more transaction fees than inflation to reward the verifier (or agent). In fact, this is a very complex subject, so I will try to keep it simple. Of course, speculative enthusiasm and currency premiums can keep a network sustainable, even if the network’s inflation is higher than the transaction fees it receives. But for a truly resilient and decentralized network, we should strive to achieve economic sustainability .

The maintenance cost (i.e. inflation) of the centralized L1s network far exceeds the transaction fee income

Let’s consider again our two favorite examples-Polygon PoS and Solana. The Polygon PoS network receives approximately US$50,000 in transaction fees per day, which translates to an annual transaction fee of US$18 million. At the same time, Polygon is allocating more than $400 million in annual inflation incentives. This resulted in an incredible 95% net loss .

As for Solana, its daily transaction fee for a long period of time was only $10,000, but with the rise of speculative frenzy, the network’s transaction fee income has significantly increased to $100,000 per day, or $36.5 million per year. But Solana is giving out an even more shocking $4 billion annual inflation reward, resulting in a net loss of 99.2% .

I collected these data from Token Terminal and Staking Rewards. I think these data are relatively conservative, and the actual data may be worse. By the way, the transaction fee that the Ethereum network gets in one day is more than the sum of the two networks in one year!

You cannot increase throughput to a level that is technically impossible

There are arguments that these networks will process more transactions and receive more fees in the future, inflation will fall, and eventually, the networks will break even. But the reality is much more complicated. First, even if we consider Solana’s lowest possible inflation rate before 2030, we will still see a 96% loss for the network. To achieve break-even, the network requires throughput far beyond the possible range: Solana needs to reach 154,000 TPS under the current transaction fees to achieve break-even, which is completely impossible considering the current hardware and bandwidth .

However, the bigger problem is that the growth of these additional transaction volumes is not without cost-they increase the demand for greater bandwidth, greater state expansion, and in general, the requirements for the system are still relatively high. Some people may further argue that there is already a lot of memory space and they can do more; but as I discussed in the technical scalability section above, this is a dubious assumption at best-given you At least 128 GB of RAM is required to synchronize a chain with only a few hundred TPS. Another view is that hardware will become cheaper-this is true, but this is not a magical solution-you either need to choose a larger scale, lower cost, or a balance of the two, and zkRollup will also benefit from Moore’s Law and Nielsen’s Law.

In the end, all centralized L1s have to increase transaction fees

In the end, there are only two solutions: a) the network becomes more centralized, and b) transaction fees become higher when the network reaches its limit. a) has its limitations, as discussed above, so b) is inevitable. You can see this on Polygon PoS, where transaction fees are slowly rising. In fact, Binance Smart Chain (BSC) has gone through this process, and it is now a sustainable network, despite the significant increase in transaction fees. Remember, we are only discussing economic sustainability here.

Before proceeding, let me point out again that there are many, many variables (such as price appreciation and volatility), and the above is definitely a simplified understanding, but I believe the overall logic will be clear.

Rollups + data sharding is more efficient and lower cost

Next is the case of Rollups + data sharding. In terms of Rollups, its maintenance cost is very , very low , the number of nodes that need to be run in a given time is very small, and it does not require expensive consensus security mechanisms, but it provides much greater throughput than any L1. Rollups can simply charge a small amount of L2 transaction fees to keep the network profitable. In terms of data availability, Ethereum is currently in a highly deflationary state. Combined with an efficient beacon chain consensus mechanism, only the lowest level of economic activity can achieve near-zero inflation.

Therefore, the entire Rollups + data sharding ecosystem can maintain sustainable development with greater scalability and potentially lower costs. In fact, becoming a zkRollups on Ethereum is the most beneficial for L1s, and I am very happy to see Solana at least considering this .

In short, the cost of Rollups + data sharding is only a small part of centralized L1, enabling it to provide orders of magnitude greater throughput at a similar cost; or provide similar throughput at a lower cost.

Short-term options

It is important to understand that the combination of Rollups + data sharding is a long-term perspective and it will take several years to mature.

But in the short term, there are two options:

  1. A sustainable centralized L1 (such as Binance Smart Chain) and Rollups network;
  2. An unsustainable centralized L1.

The first option is still too expensive for most people. Optimized Rollups such as Hermez, dYdX, or Loopring provide transaction fees similar to Binance Smart Chain (BSC), while Arbitrum One and Optimistic Ethereum networks have higher transaction fees, despite the OVM 2.0 promise that Optimistic Ethereum will release next month The network’s transaction fees will drop by 10 times.

The second option includes Polygon PoS and Solana, which currently offer lower fees, but as mentioned earlier, this is not sustainable in the long run. However, in the short term, they provide a good option for users seeking cheap deals. But wait, there is a third option: Validiums .

Validiums provides transaction fee levels similar to Polygon PoS or Solana-now the NFT minting and trading platform Immutable X has used the Validium solution to provide free NFT minting. Now, the data availability aspect of Validium can be said to be unsustainable, just like the centralized L1, although by using alternative consensus methods (such as the data availability committee), Validium is actually still cheaper. But the beauty of Validiums is that when data fragments are released, they can be directly forward compatible with Rollups or Volitions. Of course, as mentioned earlier, centralized L1s can also do the same, but this will be a more disruptive change.


  1. The blockchain industry does not yet have the technology to achieve global scale;


  2. Some centralized L1s blockchain networks offer very low transaction fees, which are actually subsidized by people’s speculation on their tokens. For users who want very cheap transaction fees, these networks are a good choice, but they need to realize that this is not a sustainable model, not to mention that they have serious decentralization and security compromises. .


  3. But if these centralized L1s networks gain user appeal, they will also be forced to increase transaction fees and will be replaced by newer and more centralized L1s. This is a race to the bottom that cannot be sustained for a long time.


  4. Currently, there are indeed sustainable network options, such as Binance Smart Chain (at least economically) or optimized Rollups, which offer a transaction fee range of approximately $0.10-$1.


  5. In the long run, Rollups + data sharding is the only solution that can scale to millions of TPS, reaching a global scale while maintaining technical and economic sustainability. They can do this while maintaining a high degree of security, decentralization, permissionless, trustless, and reliable neutrality, which is really amazing. A wise man once said: “Any sufficiently advanced technology is no different from magic.” This is the combination of Rollups and data fragments.

Finally, this is not just about Ethereum, Tezos has also turned to Rollup as the center. For all L1s, this is inevitable, that is, they either a) become a zkRollup; b) become a security and/or data availability chain for Rollups to build on; c) accept the obsolescence of its technology and completely Rely on marketing, memes and network effects.


Posted by:CoinYuppie,Reprinted with attribution to:https://coinyuppie.com/why-is-rollups-data-sharding-the-only-way-to-achieve-one-million-tps/
Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.

Like (0)
Donate Buy me a coffee Buy me a coffee
Previous 2021-09-13 08:36
Next 2021-09-13 08:37

Related articles