A modular blockchain is a blockchain designed to handle only one or two of the four core blockchain functions, rather than handling all four in a single, monolithic chain. The modular thesis argues that specialization enables each layer to scale more effectively than a jack-of-all-trades chain, and that composable “Lego blocks” of blockchain infrastructure create a better design space than monolithic L1s trying to do everything. The modular movement was popularized by Celestia’s team (especially Mustafa Al-Bassam) and has significantly influenced Ethereum’s rollup-centric roadmap.
The Four Blockchain Functions
1. Execution:
Processing transactions; running smart contract logic; computing state transitions.
- Intensive CPU/memory work
- Benefits from parallelism
- Example execution layers: Ethereum L1, Solana, rollup VMs (OP, Arbitrum, zkEVM)
2. Settlement:
Finalizing state — verifying proofs, resolving disputes, anchoring final state.
- Provides economic finality guarantees
- Settlement requires high security (Ethereum L1 is a dominant settlement layer)
- In optimistic rollups: the fraud proof process on L1 is settlement
- In ZK rollups: the validity proof verification on L1 is settlement
3. Consensus:
Agreeing on transaction order; determining which transactions happened and in what sequence.
- Prevents double-spends and forks
- Needs Byzantine fault tolerance
- Example: BFT consensus algorithms, Nakamoto consensus (PoW)
4. Data Availability:
Guaranteeing that all transaction data is published and retrievable so validators can rebuild state.
- Prerequisite for fraud proofs (auditors need the data to check for fraud)
- Most underappreciated function; historically bundled with consensus
- Dedicated DA layers: Celestia, EigenDA, Avail, Ethereum blob market (EIP-4844)
Monolithic vs. Modular
| Property | Monolithic (e.g., Solana, Ethereum L1) | Modular |
|---|---|---|
| All functions on one chain | Yes | No — distributed across layers |
| Security model | Single shared security | Composed from multiple layers |
| Scaling approach | Optimize the one chain | Scale layers independently |
| Developer UX | Familiar single chain | More complexity, more flexibility |
| Examples | Solana, BSC, early Ethereum | Ethereum + rollups + Celestia |
Monolithic chain’s bottleneck: When execution, settlement, consensus, and DA share a single resource (block space), every function competes. Making execution faster means DA constraints bind harder, etc.
Modular solution: Each layer independently scales to its optimal design point.
Ethereum as a Modular Stack
Ethereum has evolved into a de facto modular stack:
“`
┌─────────────────────────────────────────────┐
│ EXECUTION LAYER: Rollups (Arbitrum, Base, │
│ Optimism, zkSync, Starknet, Scroll…) │
├─────────────────────────────────────────────┤
│ SETTLEMENT LAYER: Ethereum L1 (fraud/ │
│ validity proofs verified here) │
├─────────────────────────────────────────────┤
│ CONSENSUS LAYER: Ethereum PoS validators │
├─────────────────────────────────────────────┤
│ DATA AVAILABILITY: Ethereum blobs (EIP-4844)│
│ OR Celestia OR EigenDA OR Avail │
└─────────────────────────────────────────────┘
“`
Pre-2023 Ethereum: All four functions handled monolithically on Ethereum L1. Expensive because DA and execution competed for the same blockspace.
Post-EIP-4844 Ethereum: DA is handled separately (blobs) — the first step toward modular DA.
Rollup-centric roadmap: Vitalik’s stated long-term Ethereum roadmap where L1 focuses on settlement + DA, and execution moves entirely to rollups.
Celestia’s Modular Vision
Celestia provides only consensus + DA:
- Does NOT run smart contracts
- Does NOT settle rollup state
- Only guarantees: transaction ordering and data availability
This allows rollup chains to:
- Use Celestia for cheap DA + ordering
- Self-settle (sovereign rollups) or settle on Ethereum
- Run any VM (EVM, SVM, MoveVM, custom)
The result: a rollup using Celestia + EigenLayer security + custom VM has assembled its own modular stack from best-of-breed components.
Data Availability Sampling (DAS)
The cryptographic technique enabling modular DA layers to work at scale:
Problem: Full block download is infeasible for light clients (blocks are too large).
DAS solution:
- Block data is erasure-coded into a larger matrix (any 50% of the matrix can reconstruct all data)
- Light nodes randomly sample tiny pieces of the matrix
- If all sampled pieces are available, with high probability the whole block is available
- Many light nodes sampling in parallel give stronger availability guarantees than a few full nodes downloading everything
DAS enables DA scaling: more light nodes = more samples = stronger DA security, while each node only downloads a tiny fraction of the block.
Current Modular Landscape (2024-2025)
DA Layers:
- Celestia (most used alt-DA)
- EigenDA (EigenLayer’s restaked Ethereum security)
- Avail (Polygon-incubated, now independent)
- Ethereum blobs (dominant by $TVL secured)
Execution Layers:
- OP Stack rollups (Base, Optimism, Zora, Mode)
- Arbitrum Orbit chains
- zkSync hyperchains
- Starknet fractal scaling
Settlement:
- Ethereum L1 (most rollups)
- Native L1 validators (sovereign rollups via Celestia)
Related Terms
Sources
Al-Bassam, M., Sonnino, A., & Buterin, V. (2018). Fraud and Data Availability Proofs: Maximising Light Client Security and Scaling Blockchains with Dishonest Majorities. arXiv:1809.09044.
Thibault, L. T., Bhatt, T., & Bhatt, A. (2022). Blockchain Scaling Using Rollups: A Comprehensive Survey. IEEE Access.
Boneh, D., Bünz, B., & Fisch, B. (2018). Batching Techniques for Accumulators with Applications to IOPs and Stateless Blockchains. CRYPTO 2019.
Ethereum Foundation. (2023). The Rollup Centric Ethereum Roadmap. Ethereum Research Blog.
Neiheiser, R., Inácio, G., Rech, L., & Montez, C. (2021). Practical Byzantine Fault Tolerance: A Survey. Frontiers in Computer Science.