The Scaling Landscape: Validium vs Rollup

Introduction: Scaling Dilemmas and the Zero Knowledge Paradigm

In the ever-evolving world of blockchain technology, scalability is a pressing concern. Among the various options available, the incorporation of zero-knowledge techniques has garnered significant interest. This exploration dives into two key players in this domain — Zero-Knowledge Rollups (ZKR) and Zero-Knowledge Validium (ZKV) — as they offer promising Layer 2 (L2) scaling solutions. Our focus here is to delve into the nuances of these alternatives and address the crucial question of what components should remain on Layer 1 (L1).

Sequential vs. Parallel Throughput: A Pivotal Distinction

Consider a scenario where two users are interacting with the same data, e.g. a decentralized exchange (DEX). The nature of transactions depending on each other’s output state reveals two distinct options:

1. Sequential Throughput: Users are aware of each other’s output state. This involves a sequential process where one user completes, signs, and submits their transaction to the L2. The second user then pulls the output state before constructing, signing, and submitting their transaction. However, limitations in signing time and data transmission result in a limited sequential throughput. As an order of magnitude estimate, we would approximate around one sequential transaction per second.
2. Non-Aware Sequential Throughput: Users lack full awareness of each other’s output state. This introduces uncertainty and potential complications. To illustrate, in a DEX scenario, this option exposes users to potential manipulation, allowing the centralized L2 to determine the transaction order — a phenomenon akin to sandwich attacks. Even in a trusted, decentralized L2 environment, the uncertainty introduced to transactions remains a considerable drawback.

Our proposed solution consists of ensuring the independence of all L2 transactions within the same batch proof in order to mitigate complications. However, this decision comes with a trade-off, as it restricts the sequential throughput of the system.

Managing Extensive Data for Scalability

While the overarching goal is to achieve substantial throughput, the emphasis lies on increasing parallel throughput. This necessitates the parallelization of transactions, each interacting with a different part of the blockchain. Therefore, significant data storage will become a requirement for the L2, including many wallets and concepts such as sharded L2 smart contracts.

Given the data storage requirements we will run into at the Casper Association, storing this data directly on the L1 might not be possible. Many blockchains don’t support a single smart contract containing this much data. Additionally, the cost of state updates in terms of gas fees poses a challenge, potentially increasing the cost of using the L2. This would diminish one of the primary benefits of ZK-based L2 solutions.

Conclusion: Charting the Course for a Comprehensive Ecosystem

In the pursuit of a scalable blockchain ecosystem capable of supporting both transactions, a digital art ecosystem, and a myriad of L2 dApps, we must evaluate the different scaling solutions available. As the landscape continues to evolve, reconsideration must be kept open, but within the current climate we can conclude that the Casper Association and its initial Kairos project will employ a Validium rather than a Rollup.