Predictable Performance of DApps: From Appchain to Elastic Block Space - From Artela Builder

Author: Web3Pignard

"Predictable performance" is an inevitable development path for DApp support in large-scale adoption. Appchain and elastic block space are two different solutions.

Artela Whitepaper

On June 20, the innovative parallel EVM Layer 1 project Artela released its whitepaper on 'Full-Stack Parallelization,' aiming to fully unlock blockchain scalability and enable DApps to have 'predictable performance.'

Predictable performance refers to providing DApps with predictable TPS, which is crucial for DApps in certain application scenarios. DApps deployed on public chains generally must compete for blockchain computing power and storage space with other DApps. In cases of network congestion, this leads to high transaction execution costs and delays, severely limiting the rapid development of DApps. Imagine if users of a decentralized instant messaging app were unable to send and receive messages because the underlying blockchain network's block space was occupied by other DApps. This would be disastrous for user experience.

To solve the problem of "predictable performance," the most common approach is to use an application-specific blockchain (Appchain), which is a blockchain dedicated to a specific application.

Artela innovatively proposed the Elastic Block Space (EBS) solution, based on the concept of elastic computing. It dynamically adjusts block resources from the protocol level according to the specific needs of DApps, providing independent scalable block space for high-demand DApps.

This article will introduce both appchains and elastic block space, comparing their pros and cons.

The Path of Appchains

Appchains are blockchains created to run a single DApp. Application developers build a new blockchain from scratch with a custom virtual machine to execute transactions from user interactions with the application, rather than building on an existing blockchain. Developers can also customize different elements of the blockchain network stack—consensus, networking, and execution—to meet specific design requirements, addressing issues like high congestion, high costs, and fixed features on shared networks.

The concept of appchains is not new: Bitcoin can be seen as an appchain for "digital gold," Arweave as an appchain for permanent storage, and Celestia as an appchain for providing data availability.

Since 2016, appchains have included not only monolithic blockchains but also multi-chain ecosystems, such as Cosmos and Polkadot. Cosmos envisioned a world of interconnected blockchains, solving cross-chain interaction issues with its IBC protocol, allowing seamless blockchain interaction. Polkadot aims to be a perfect blockchain scaling solution, with its chains called parachains, emphasizing shared security and cross-consensus communication.

By late 2020, as Ethereum's scaling research focused on sidechains, subnets, and Layer 2 Rollups, appchains also evolved. Sidechains like Polygon, and subnets like Avalanche improved experience and performance, enhancing overall service capability. Layer 2 Rollups, such as OP Stack and Polygon CDK, aim to increase Ethereum's throughput and scalability to meet growing transaction demands while providing extensive interoperability.

Many applications have been built on appchains across various ecosystems. For example, Axie launched its Ethereum sidechain Ronin in early 2021; DeFi Kingdoms migrated from Harmony to Avalanche subnets in late 2021; Injective launched its DeFi appchain using Cosmos SDK in November 2021; dYdX announced in mid-2022 that its V4 version would use Cosmos SDK for a standalone appchain; Uptick Network launched its Uptick Chain in 2023 to support Web3 ecosystem applications, with rich commercial protocol layers in its infrastructure.

Advantages and Disadvantages of AppChains

Appchains have full sovereignty over their blockchains, rather than relying on underlying Layer 1. This is a double-edged sword.

Pros:

• Sovereignty: Appchains can solve issues through their governance schemes, maintaining the independence and autonomy of individual application projects, preventing various interferences.

• Performance: They can meet the low-latency and high-throughput needs of applications, providing users with a good experience and greatly improving the operational efficiency of DApps.

• Customizability: DApp developers can customize the chain according to their needs, even creating an ecosystem, offering flexible evolution.

Cons:

• Security Issues: Appchains are responsible for their security, balancing node quantity, maintaining consensus mechanisms, and avoiding staking risks, making the network relatively insecure.

• Cross-Chain Issues: As independent chains, appchains lack interoperability with other chains (applications), facing cross-chain problems. Integrating cross-chain protocols increases cross-chain risks.

• Cost Issues: Building additional infrastructure for appchains involves significant costs and engineering time. Additionally, there are costs for running and maintaining nodes.

For startups, the disadvantages of appchains significantly impact the operation of market-entry DApps. Most startup development teams cannot effectively solve security and cross-chain issues and are discouraged by the high costs in human resources, time, and money. However, predictable performance is a rigid demand for specific DApps. Therefore, the market urgently needs a Layer 1 solution for predictable performance.

Elastic Block Space

In Web2, elastic computing is a common cloud computing model that allows systems to dynamically expand or shrink computing, memory, and storage resources as needed to meet changing demands without worrying about capacity planning and engineering design for peak usage.

Elastic block space adjusts the number of transactions that blocks can accommodate based on network congestion. For specific application transactions, the blockchain network provides stable block space and TPS guarantees through elastic computing, achieving "predictable performance."

MegaETH proposed a similar concept of "elastic dynamic scaling," believing it to be the inevitable development path for large-scale DApp adoption, predicting the following technical developments in the next 1-3 years:

• Phase One: Horizontal scaling at the validator node level.

• Phase Two: Static scaling at the chain level.

• Phase Three: Dynamic horizontal scaling at the chain level.

Artela has truly realized this concept, solving the core issue of "how to coordinate horizontal scaling of validator nodes to support elastic computing" in the first phase. As the protocol in the Artela network grows, it can subscribe to elastic block space to handle the growth of protocol users and throughput. Elastic block space provides independent block space for DApps with high transaction throughput demands, allowing them to scale with growth. Essentially, block space determines the data capacity each block can store, directly affecting transaction throughput. When DApps experience surges in transaction demand, subscribing to elastic block space becomes useful for efficiently handling the increased load without impacting the underlying blockchain.

The implementation of elastic computing can be divided into "real-time elasticity" and "non-real-time elasticity." "Real-time elasticity" typically refers to minute-level response scaling, while "non-real-time elasticity" requires response scaling within a defined time. Artela adopts the "non-real-time elasticity" approach. When the network detects the need for scaling, it initiates a scaling proposal, and after one or more epochs (rather than real-time), the entire network's validator nodes complete the scaling and submit proof of scaling for other validators to challenge.

Artela's elastic block space solution draws heavily from distributed database concepts and continues blockchain sharding technology. From the perspective of "computational sharding," it scales based on application traffic demand, avoiding "cross-shard transactions" and maintaining developer and user experience. At the same time, adopting the relatively easier-to-implement "non-real-time elasticity" meets the actual needs of many DApps, enhancing applicability.

It is worth mentioning that as a solution for horizontally scaling blockchain performance, the prerequisite for elastic block space is "parallel transactions." Only when transaction parallelism is achieved does it become necessary to horizontally expand node resources to improve transaction throughput.

For Layer 1 blockchains like Ethereum, serial transaction processing is the most direct performance bottleneck, with block size limited by variable block gas limits (up to 30,000,000 gas). Thus, Layer 2 scaling solutions are sought.

For high-performance Layer 1 blockchains like Solana, which support parallel transaction execution and can horizontally scale performance, the problem of "predictable performance" during peak demand for DApps remains unresolved. Solana's "local fee markets" solution aims to prevent any single demand from monopolizing scarce block space, limiting time-sensitive fee spikes and mitigating the negative impacts of sudden demand peaks. For example, during NFT launches, NFT issuers quickly consume the compute unit (CU) limits for each account, requiring subsequent transactions to increase priority fees to be processed within the limited space of that account.

Artela's elastic block space solution, addressing surges in transaction demand, extends Solana's "local fee markets" concept, ensuring "predictable performance" for DApps and preventing network-wide fee spikes and congestion, achieving a dual benefit.

Conclusion

Whether it is appchains or elastic block space, the essence is to solve the problem of different DApp performance demands on the blockchain, or "predictable performance." These solutions are not about being good or bad but about suitability. These solutions remind the author of the "Fat Protocol Theory"—proposed by Joel Monegro in 2016, focusing on "how crypto protocols should capture more value (compared to the collective value captured by applications built on top of them)."

Appchains are essentially thin protocols. Especially when Layer 1 adopts a modular architecture, the protocol layer is entirely customized by the application layer, bringing better value accumulation mechanisms for applications but also high costs and limited security.

Elastic block space is essentially a fat protocol, an extended feature of the underlying Layer 1 protocol, effectively lowering the entry barrier for participants needing "predictable performance" while the protocol can also capture application value, generating a positive feedback loop.