The Role of Shared Sequencers

The dominant types of Rollups 📜 on Ethereum are Optimistic, ZK and Validium, off-loading execution from L1 with Validium having data availability off-chain. Each type of Rollup comes with its own trade-offs. However, regardless of the Rollup type, some core components remain the same (see figure below), including the sequencer. It performs the most important functions in the entire Rollup system, including rollup transactions ordering, batching (many transactions put in a bundle) and validating (executing transaction and updating rollup state). In other words, without the well designed and functioning sequencers, there will be no Rollups. Because of its critical role, its earnings is also significant from users’ L2 fee spending. Below is sequencer fees as of late May this year (@ $3000/ETH):

• 🛡️Arbitrum: 58,573.20 ETH ($175,719,600)
• 🔵Base: 12,277.61 ETH ($36,832,830)
• 🔄zkSync: 39,976.50 ETH ($119,929,500)

However, most L2s currently use centralized sequencers (just one single server to handle the whole rollup transactions) for better throughput, low latency and instant finality. This approach comes with significant drawbacks, risks and trust assumptions. For instance, Optimism’s fault proof system will only be active after the Bedrock upgrade. Before the upgrade, users can only trust the single sequencer (one server...) with no transaction validity guarantee and hoping the server doesn’t go down 💔

Given the crucial role of a sequencer in Rollups, I shall point out some of the key challenges with the centralized sequencer set up:

• 🚥Liveness & Neutrality - Vulnerable single point of failure, sacrificing resilience and neutrality.
• ♟️Monopoly- Centralized Sequencers may engage in price gouging and monopolistic practices
• 🏊🏻‍♀️Fair Ordering and Private Mempools - Centralized Sequencers can prioritize public mempools for higher fees and may ignore private mempools, acting as rational actors.
• 🌉Fragmentation Across Rollups - Fragmentation in cross-rollup bridging and messaging due to independent sequencers that only communicate with L1, not other rollups.

To solve the challenges mentioned above, the Shared Sequencer approach is proposed, removing the need for rollups to create and operate their own sequencers. It is a decentralized network of nodes (with consensus and leader rotation) that sits between L1 and Rollups as a middleware layer (see figure below). In addition to being rollup agnostic and censorship resistant with stronger economic security, it enables atomic cross-rollup composability, unlocking valuable use cases such as conditional cross rollup transactions (something omnichain bridges aim to provide). Below is key solutions provided by shared sequencers:

• 🚥Liveness & Neutrality - Decentralization lowers vulnerability and promotes neutrality, reducing the impact of price discrimination.
• ♟️Monopoly- Substantially more sequencers in the decentralized network to counter monopoly versus relying on the goodwill of any one actor.
• 🏊🏻‍♀️Fair Ordering and Private Mempools - Decentralization lowers MEV impact, enabling fair ordering (relying on an honest majority) and allowing private mempools to function effectively.
• 🌉Fragmentation Across Rollups - Shared sequencers reduce rollup silos as they use a common sequencing network layer to enable cross-rollup interoperability.