Glossary · Scaling

Modular blockchain

Published 22 May 2025 · Updated 2 Nov 2026 · 14 min read

A modular blockchain is a design that splits execution, settlement, consensus and data availability across specialized layers instead of one chain. Monolithic blockchains keep those functions together in a single base layer. An Ethereum rollup separates execution from its base layer, while Solana represents the monolithic approach.

  • Data availability is the function most modular designs move to another layer or share.
  • Publishing transaction data outside Ethereum adds a separate trust assumption.
  • On 2 November 2026, most value tracked by L2BEAT uses chains publishing data to Ethereum.
How a modular blockchain splits workRollups and other modular designsThe data availability problemData availability layers in November 2026What each node has to downloadFrameworks for building modular chainsThe trade-offs of modular designsHow modular blockchains developedDecisions behind a modular chainModular blockchain FAQ

How a modular blockchain splits work

Every chain must run transactions, check results and settle disputes, agree on ordering, and publish data for others to check. These jobs are execution, settlement, consensus and data availability. A monolithic blockchain performs all four in its base layer. Ethereum rollups move transaction processing outside that layer while retaining base-layer functions.

FunctionWhat it doesMonolithic chain such as SolanaEthereum rollupChain that publishes to Celestia
ExecutionWhat it doesRuns transactions and updates the chain's stateMonolithic chain such as SolanaThe chain's own nodesEthereum rollupThe rollup execution layerChain that publishes to CelestiaThe chain's execution layer above Celestia
SettlementWhat it doesVerifies proofs, resolves fraud disputes and supports bridgingMonolithic chain such as SolanaThe same base layerEthereum rollupContracts on EthereumChain that publishes to CelestiaA separate settlement layer, if used
ConsensusWhat it doesAgrees on transaction ordering and establishes finalityMonolithic chain such as SolanaThe chain's validator setEthereum rollupEthereum orders posted data, while the sequencer orders transactions inside the rollupChain that publishes to CelestiaCelestia orders posted data, while the sequencer orders transactions inside the rollup
Data availabilityWhat it doesPublishes transaction data so network participants can download itMonolithic chain such as SolanaThe chain's own networkEthereum rollupEthereum blobs or calldataChain that publishes to CelestiaCelestia's data availability layer

Celestia argues that one base layer cannot be optimized for all four jobs. A rollup execution layer can share settlement and publication with other modular chains. Ethereum's October 2020 rollup-centric roadmap prioritizes blockchain consensus and data availability at the base layer. In May 2021, Vitalik Buterin emphasized that regular users must still be able to run nodes. Splitting work does not remove that constraint.

A modular blockchain changes where users place trust rather than removing it. Execution, settlement and publication depend on different checks, and a weak dependency can limit the whole rollup. This does not solve the blockchain trilemma without trade-offs. In November 2026, L2BEAT's glossary does not count a project as a rollup when its data is stored off a permissionless L1. Outside publication adds an assumption about another provider.

Rollups and other modular designs

Two choices distinguish most modular blockchain designs: who checks the chain's state and where transaction data is published. A rollup framework can support different arrangements. Layer 2 designs therefore need more description than a framework name alone.

Figure 1 · L2BEAT categories, read on 2 November 2026Build a modular stackChoose state verification and data publication to see the category, trust assumptions, withholding risks and real project examples.
Who checks the state
Where data is published
ResultL2BEAT: RollupsL2BEAT: OthersL2BEAT: Validiums & OptimiumsL2BEAT: Validiums & OptimiumsL2BEAT: OthersL2BEAT: OthersL2BEAT: RollupsL2BEAT: OthersL2BEAT: Validiums & OptimiumsL2BEAT: Validiums & OptimiumsL2BEAT: OthersL2BEAT: OthersNot listed on L2BEATNot listed on L2BEATNot listed on L2BEATNot listed on L2BEATNot modularOptimistic rollupEthereum orders and publishes the data, and its contract settles the results.Optimium design, listed as OthersNo DA bridge proves to Ethereum that the data was published on Celestia.OptimiumOnly with a DACert Verifier that checks EigenDA certificates. Without one, L2BEAT lists the project in Others.OptimiumThe Vector bridge posts Avail data roots to Ethereum, so the contract can check them.Optimium design, listed as OthersL2BEAT lists all 17 committee projects in Others, for committee rules, proof gaps or a missing DA bridge.Chain with its own data layerAn Ethereum contract settles, but the chain's validator set keeps the data. L2BEAT counts this as a custom DA solution.ZK rollupEthereum orders and publishes the data, and its contract settles the results.Validium design, listed as OthersNo DA bridge proves to Ethereum that the data was published on Celestia.ValidiumOnly with a DACert Verifier that checks EigenDA certificates. Without one, L2BEAT lists the project in Others.ValidiumThe Vector bridge posts Avail data roots to Ethereum, so the contract can check them.Validium design, listed as OthersL2BEAT lists all 17 committee projects in Others, for committee rules, proof gaps or a missing DA bridge.Chain with its own data layerAn Ethereum contract settles, but the chain's validator set keeps the data. L2BEAT counts this as a custom DA solution.Sovereign rollupIts own nodes check the state. Ethereum only orders and publishes the data.Sovereign rollupIts own nodes check the state. Celestia only orders and publishes the data.Sovereign rollupIts own nodes check the state. EigenDA stores and serves the data but does not order it.Sovereign rollupIts own nodes check the state. Avail only orders and publishes the data.Monolithic chainOne chain executes, settles, orders and publishes its own data.Trusted for dataEthereum's validators. Full nodes sample blob data with PeerDAS, and blobs stay available for about 18 days.Celestia's validators, plus enough light nodes sampling each block and one honest bridge node to rebuild it.EigenDA operators backed by restaked ETH and EIGEN. Safety holds while an attacker has less than 33% of a quorum's stake.Avail's validators and sampling light clients. The Vector bridge posts data roots to Ethereum about every 2 hours.The committee members who sign that they hold the data. AnyTrust assumes at least two of them are honest.The chain's own validator set.Trusted for stateAt least one honest node that can post a fraud proof during the challenge period.The proof system and the verifier contract that checks every validity proof.The chain's own nodes. They decide which chain is valid, and the data layer does not.Its validators, and every full node that re-executes the blocks.If data is withheldEthereum's consensus guarantees the data was published, so anyone can rebuild the chain and exit through the contract.A validity proof still blocks theft, but without the data users can have funds frozen and withdrawals restricted.Without the data nobody can build a fraud proof. L2BEAT: funds can be lost if the data is not available on the external provider.L2BEAT: a malicious sequencer can collude with the proposer to finalize an unavailable state, which can cause loss of funds.If enough committee members sign but withhold the data, the proposer can finalize an unavailable state, and funds can be lost.The validator set that keeps the data can withhold it, and the contract on Ethereum cannot check.The rollup's nodes stop at the missing data. Bridges to the chain rely on whoever reports its state.Full nodes refuse a block whose data they cannot download, because they download and execute everything themselves.
On L2BEAT, 2 November 202643 projects post data to Ethereum. 21 of them are rollups on L2BEAT, 10 optimistic.Examples with fraud proofs: Arbitrum One, Base, OP Mainnet43 projects post data to Ethereum. 21 of them are rollups on L2BEAT, 11 ZK.Examples with validity proofs: Scroll, Starknet, ZKsync Era6 projects on L2BEAT, all in Others with no DA bridgeEclipse, LightLink, Manta Pacific, Nillion, Orderly, Towns7 projects on L2BEAT. Celo and RISE use a DACert Verifier, but RISE stays in Others.Aevo, Celo, Fuel Ignition, MegaETH, RISE, Ronin, Soon2 projects on L2BEAT, both with the Vector bridgeLens, Sophon17 projects on L2BEAT, 11 of them with a single member. None qualifies as a validium or optimium.A committee counts only with at least five outside members and a high enough threshold4 projects on L2BEAT with a custom DA solution, all in OthersFraxtal (team-run), Gnosis Chain, Hyperliquid, Polygon PoS (104 validators)L2BEAT does not list sovereign rollupsEvolve, formerly Rollkit, launches such chains on CelestiaNot a modular designSolana is the usual example of a chain that does all four functionsShare of all value on L2BEAT secured with this data layer73.5%0.24%1.3%0.01%0.35%24.1%Not tracked
All six tracked Celestia users lack a DA bridge. L2BEAT also lists all 17 committee projects in Others.A sovereign rollup uses its own nodes to decide the valid chain while publishing data on another layer.Validity proofs plus outside data and a DA bridge describe validiums such as Sophon and Lens on Avail.Polygon PoS uses its own validator set for data and appears in Others, despite an Ethereum contract.Fraud proofs plus outside data and a DA bridge describe an optimium, such as Celo on EigenDA.Ethereum contracts and data define a rollup. An optimistic rollup uses fraud proofs, while a ZK rollup uses validity proofs.A monolithic chain such as Solana performs all four functions within one base consensus layer.

Categories follow L2BEAT definitions. Projects and counts were read on L2BEAT on 2 November 2026.

An optimistic rollup and a validium design differ in both proof type and publication assumptions. Sidechain designs use their own validators and a two-way bridge. They do not post state changes and transaction data back to Ethereum. L2BEAT's 2 November 2026 classification lists 22 rollups, three validiums and optimiums, and 69 projects in Others. Posting data to the base chain alone does not establish rollup status. Proof systems and independent challengers also affect rollup classification.

The data availability problem

A producer could publish a block's summary but hide its transactions. Other participants could not check the result or rebuild the chain state. This is data withholding. Data availability means the block's data was published and can be downloaded. Monolithic full nodes download and execute every transaction. A light node, a type of light client, needs another way to check publication.

Al-Bassam, Sonnino and Buterin described data availability sampling in September 2018. In November 2026, Celestia applies Reed-Solomon erasure coding, extending k-by-k data into a 2k-by-2k square. The block header holds one root of all row and column Merkle roots. By default, each light node requests 16 random shares per block with Merkle proofs. Enough sampling lets honest bridge nodes rebuild blocks. Fraud proofs reject bad encoding. This needs enough light nodes and one honest bridge connection per sampler. A namespace lets each rollup retrieve only its own data.

Ethereum introduced blobs through EIP-4844 in March 2024. In November 2026, each blob holds 128 KiB and remains available for about 18 days. Contracts check commitments using BLOBHASH and a KZG point-evaluation precompile. Since Fusaka in December 2025, PeerDAS lets full nodes sample and download about one eighth of blob data. Publication is not permanent storage. A rollup still needs its own history storage after the publication window closes.

Data availability layers in November 2026

A modular blockchain can publish data to its settlement chain, an external network or a committee. The comparison below reflects 2 November 2026. Capacity figures describe different mechanisms, so throughput alone does not explain how publication is checked.

LayerLive sinceCapacityWhom users trustHow publication is checked
Ethereum blobsLive sinceMarch 2024Capacity12-second blocks, target 14 blobs, maximum 21Whom users trustUsers rely on Ethereum validators without an external committee honesty assumptionHow publication is checkedContracts check commitments through BLOBHASH and the KZG point-evaluation precompile
CelestiaLive sinceOctober 2023CapacityAbout 3-second blocks, configured maximum 32 MiBWhom users trustUsers rely on validators plus light-node sampling, or Fibre signatures from at least two-thirds of voting powerHow publication is checkedLight nodes sample shares. Blobstream, its bridge to Ethereum, is legacy since September 2026
EigenDALive sinceApril 2024CapacityV2 states 100 MB per secondWhom users trustUsers rely on quorums secured through ETH and EIGEN restakingHow publication is checkedContracts verify quorum certificates with DACert Verifier where deployed
AvailLive sinceJuly 2024Capacity20-second blocks, default size 5 MiBWhom users trustUsers rely on Avail consensus and data availability samplingHow publication is checkedVector posts data roots to Ethereum every 360 blocks
Data availability committeeLive sinceDeployment dates varyCapacityCapacity depends on deploymentWhom users trustUsers rely on committee members that sign for held dataHow publication is checkedContracts check committee signatures rather than sampling published data
Figure 2 · Page team tally of L2BEAT, 2 November 2026Projects and value by layerCompare project shares with secured value, switch to L2s only, and see which projects sit behind each layer.
View
Share of projects and of value securedShare of the 94 projectsShare of value secured
Projects in this view43 of 9446% of active projects on L2BEAT
Share of value in this view73.5%Of the value secured by all 94 projects
Total value secured$44.95 billionAll 94 projects, L3s included
Projects in this view4 of 944% of active projects on L2BEAT
Share of value in this view24.1%Of the value secured by all 94 projects
Total value secured$44.95 billionAll 94 projects, L3s included
Projects in this view15 of 9416% of active projects on L2BEAT
Share of value in this view1.6%Of the value secured by all 94 projects
Total value secured$44.95 billionAll 94 projects, L3s included
Projects in this view17 of 9418% of active projects on L2BEAT
Share of value in this view0.35%Of the value secured by all 94 projects
Total value secured$44.95 billionAll 94 projects, L3s included
Projects in this view43 of 9446% of active projects on L2BEAT
Share of value in this view86.4%Of the value secured by L2s
Total value secured$38.23 billionL2s only, without L3s

Ethereum serves 46% of projects and secures 86.4% of L2-only value, compared with 73.5% including L3s.

Four projects with custom DA solutions account for 24.1% of all tracked value, despite representing only 4% of projects.

EigenDA secures $594 million, Celestia $107 million and Avail $6 million. Their 15 projects represent under 2% of value.

Committees serve 18% of projects and secure $159 million, or 0.35% of value. Eleven of 17 committees have one member.

The L2-only view removes L3s, including host-chain projects, raising Ethereum's value share to 86.4%.

The page team tallied 94 active L2BEAT projects securing $44.95 billion, using figures read on 2 November 2026.

Value secured is not a quality score. Value on custom DA solutions in November 2026 mostly comes from Hyperliquid, an L3, and Polygon PoS. Blobstream is legacy since September 2026. L2BEAT says no tracked Celestia user verifies data through it. Arbitrum Nova moved from AnyTrust to blobs on 31 August 2026 and is now classified as an optimistic rollup. In late October 2026, Ethereum blob use averaged 35% of the target, rather than the available maximum.

What each node has to download

A node that cannot keep up must depend on other participants. A monolithic chain asks validators to process the whole chain's transactions. A modular blockchain can let a light node check publication with a much smaller download. A rollup full node still has a different job from a sampler. Comparing their bandwidth figures shows workload differences, not equivalent security roles.

Figure 3 · Published requirements and page team calculationsCompare node download requirementsSwitch minimum and recommended rates, then change Celestia's square size to compare full processing with sampled downloads on a logarithmic ladder.
Focus
Figures
16 by 16
Download rate, log scalePublished requirementPage team calculation
Selected2 Gbit/sSolana staked validator
Against a Celestia light node35,714×Times the 56 Kbps in Celestia docs
Light node download, calculated59.6 kbit/sSquare 16 by 16, one block every 2.84 s

Anza's 2026 requirements specify at least 2 Gbit/s symmetric bandwidth for a staked Solana validator, with 10 Gbit/s recommended.

Ethereum's published full-node minimum is 25 Mbit/s. Recommended bandwidth is 50 Mbit/s down and 15 Mbit/s up.

Since December 2025, PeerDAS lets Ethereum full nodes download about one eighth of blob data, approximately 229 kbit/s at maximum capacity.

Celestia's documentation specifies 56 Kbps for a light node, compared with 1 Gbps for validators and bridge nodes.

Increasing the square from 4 to 256 raises calculated sampling traffic from about 39 to 319 kbit/s.

Requirements come from Anza, ethereum.org and Celestia. The page team's light-node, PeerDAS and Base calculations reflect 2 November 2026.

Published minimums and recommendations are not measured traffic. The page team's sampling and PeerDAS estimates exclude network overhead, signatures and other header data. On 2 November 2026, Base posted 1.09 GiB of data to Ethereum in the past day. A Base full node therefore reads about 108 kbit/s of its own data on average. Light nodes still depend on sufficient sampling and honest connections. Someone must store and serve full data, even when a modular design reduces what individual users download.

Frameworks for building modular chains

The framework shapes a modular blockchain's execution environment, but it does not alone determine settlement or publication guarantees. An appchain design also needs decisions about validators, bridges and data. These framework counts reflect 2 November 2026 and use different scopes.

FrameworkWhat it buildsWhere chains publish dataStatus in 2026
OP StackWhat it buildsRollup chains and alternative-data chains using a shared execution frameworkWhere chains publish dataEthereum or alternative data availability layersStatus in 202637 active L2BEAT projects, 31 Superchain mainnets on 2 November 2026
ArbitrumWhat it buildsRollup chains and chains using AnyTrust committee publicationWhere chains publish dataEthereum, an AnyTrust committee or an Alt-DA layerStatus in 202618 active projects on L2BEAT on 2 November 2026
ZK StackWhat it buildsZK rollup and other validity-proof chains in the ZKsync networkWhere chains publish dataEthereum, or an outside layer such as AvailStatus in 20267 active L2BEAT projects, 18 chains listed by ZKsync on 2 November 2026
Agglayer CDKWhat it buildsChains connected through Agglayer and its pessimistic proof systemWhere chains publish dataVaries by chainStatus in 20268 active projects on L2BEAT on 2 November 2026
Cosmos SDKWhat it buildsSovereign application-specific chains that connect to other chains through IBCWhere chains publish dataEach chain publishes its own data, including Celestia, built with the Cosmos SDKStatus in 2026More than 200 production chains on 2 November 2026
Polkadot SDKWhat it buildsParachains that inherit security from Polkadot's validator setWhere chains publish dataPublication depends on the deployed chain's configurationStatus in 202687 registered paras on 2 November 2026
EvolveWhat it buildsSingle-sequencer chains that publish their transaction data to CelestiaWhere chains publish dataCelestia as the publication layerStatus in 2026Single sequencer, execution interface and connection to Celestia in November 2026

Ecosystem labels change without changing every component underneath. Base left the Superchain on 4 March 2026, although L2BEAT still identifies its OP Stack foundation in November 2026. Astria's shared sequencer network halted on 1 December 2025. In November 2026, Espresso describes itself as a settlement layer, while Rollkit is now Evolve. Agglayer's pessimistic proof system limits a compromised chain to its own deposits. Shared branding does not replace checking each modular deployment.

The trade-offs of modular designs

Separating functions moves responsibilities between systems rather than eliminating them. A modular blockchain needs explicit answers about missing data, unavailable providers and bridges. A monolithic design keeps more work together, but modular designs expose these dependencies across separate layers.

Withheld transaction data

A validity proof cannot restore missing transaction data. In a validium, withholding can freeze funds and restrict withdrawals, even when the state transition is valid.

Missing publication checks

Without a DA bridge, the sequencer is trusted to report data roots honestly. Collusion with the proposer can finalize unavailable state and cause losses.

Small signing committees

A small data availability committee concentrates trust in fewer signers. AnyTrust assumes at least two of N members are honest, not merely that signatures exist.

Outside layer outages

If external publication stops, a chain may stop progressing. Some projects fall back to Ethereum. AnyTrust switches to rollup mode when signatures do not arrive.

Expiring published data

Ethereum blobs last about 18 days in November 2026. Celestia's Fibre retains payloads four hours by default, without sampling. Each rollup must preserve its history.

Fragmented liquidity and bridges

Separate execution layers fragment liquidity rather than sharing one state. A cross-chain bridge adds another dependency when users move assets between modular chains.

How modular blockchains developed

The architecture emerged through research into checking data without downloading whole blocks, then through networks and upgrades that put it into practice. These milestones trace modular designs from early sampling research to changes in where an active chain publishes data.

  1. September 2018

    Sampling enters availability research

    Mustafa Al-Bassam, Alberto Sonnino and Vitalik Buterin publish fraud and data availability research, describing probabilistic checks for light clients.

  2. May 2019

    LazyLedger separates execution from publication

    Mustafa Al-Bassam proposes a blockchain focused on ordering and publishing transaction data. Execution stays with the clients that need it.

  3. October 2020

    Ethereum's rollup-centric roadmap takes shape

    Vitalik Buterin's rollup-centric roadmap directs Ethereum's base layer toward consensus and data availability, with transaction execution moving to rollups.

  4. October 2023

    Celestia starts Mainnet Beta

    Celestia starts Mainnet Beta on 31 October 2023, providing a base layer for consensus and data publication rather than transaction execution.

  5. March 2024

    Ethereum adds temporary data blobs

    Dencun introduces Ethereum blobs on 13 March 2024. EIP-4844 gives rollups a publication format whose data is retained temporarily.

  6. December 2025

    PeerDAS brings sampled blob downloads

    Fusaka introduces PeerDAS on 3 December 2025, allowing Ethereum full nodes to sample blob data rather than download all of it.

  7. July 2026

    Celestia shortens its block interval

    Celestia cuts block time to about three seconds on 1 July 2026. The change increases publication capacity at the configured block size.

  8. August 2026

    Nova changes its publication layer

    Arbitrum Nova replaces its AnyTrust committee with Ethereum blobs on 31 August 2026. L2BEAT reclassifies the chain as an optimistic rollup.

Decisions behind a modular chain

We define the responsibilities before choosing components: execution, settlement, consensus and data publication. A modular blockchain specification should identify which system checks each promise and what happens when that system stops. Our Substrate development service covers custom parachains and cross-chain bridges. Founded in 2007, PixelPlex has built blockchain systems since 2013.

PixelPlex case icon

Echo has been live since 2017. Our Echo protocol case study describes its own consensus and small block-producer committee. Two-way pegs connect Echo to Ethereum and Bitcoin. Its ZK-rollup bridge executes transactions on Echo and posts a validity proof on Layer 1.

  1. Execution and framework

    Specify transaction processing and the framework separately from settlement. Record which execution responsibilities stay on the chain and which move elsewhere.

  2. Settlement and proofs

    Define the settlement contract, proof checks and dispute process. We offer ZK rollup development for teams planning validity-proof-based execution.

  3. Publication and verification

    Choose where transaction data is published and how a DA bridge checks publication. Document the trust assumption when no bridge exists.

  4. Sequencing and bridges

    Identify who orders transactions and which bridges connect execution layers. Keep consensus responsibilities distinct from the settlement contracts that verify results.

  5. Nodes and history

    Specify node roles, bandwidth assumptions and retention responsibilities. Plan how the chain's history remains available after temporary publication windows expire.

Discuss the boundaries of your chain design

We can discuss execution, proof checks and publication requirements against the scope of your planned chain.

Discuss your design

Modular blockchain FAQ

What is a modular blockchain in simple terms?

A modular blockchain is a chain design that assigns different jobs to different layers instead of keeping everything in one network. It separates some combination of execution, settlement, consensus and publication. The important detail is which layer checks each responsibility.

How do modular and monolithic blockchains differ?

A monolithic blockchain puts all four functions in one base layer, while a modular design distributes responsibilities. The difference concerns architecture, not a promise about speed or safety. Separate execution layers can share settlement and publication without becoming one execution environment.

Is Ethereum itself a modular blockchain?

Ethereum's base layer does all four functions for its own transactions. For rollups, it also acts as the settlement and data availability layer. Ethereum with its rollups therefore forms a modular stack. A bridge alone does not give a connected chain Ethereum's guarantees.

What does data availability mean for rollup users?

Data availability means participants can obtain the transaction data needed to verify a block or reconstruct state. It is not the same as permanent archival storage. Sampling checks publication probabilistically, so a light node can check availability without downloading the entire block.

What does Celestia do in a modular stack?

Celestia supplies consensus and data availability, not settlement or transaction execution for the chains publishing to it. Each chain must arrange those other functions separately. Its November 2026 documentation still describes Mainnet Beta as experimental.

Are rollups a type of modular blockchain?

A rollup is a modular design because execution sits outside the layer supplying consensus and publication. A sovereign rollup differs in who decides validity: its own nodes, rather than a settlement contract. L2BEAT does not list sovereign rollups in November 2026.

Are modular blockchains less secure than monolithic chains?

Modular blockchains do not share a single security level. A modular blockchain using outside publication depends on that provider as well as its proof system. Data checks, bridge design and committee thresholds matter. A validity proof does not by itself guarantee retrievable data.

Related terms

All definitions
Sources (35)

Get updates about blockchain, technologies and our company

We will process the personal data you provide in accordance with our Privacy policy. You can unsubscribe or change your preferences at any time by clicking the link in any email.

Follow us on social networks and don't miss the latest tech news

  • facebook
  • X
  • linkedin
  • instagram
Stay tuned and add value to your feed