Glossary · Blockchain basics

Distributed ledger

Published 28 Apr 2025 · Updated 3 Nov 2026 · 15 min read

A distributed ledger is a transaction record shared across several nodes whose copies stay in step through a consensus mechanism. No single database has to hold the only official version. Some nodes keep only records that concern them. Distributed ledger technology, or DLT, enables this shared recordkeeping.

  • Blockchain is one kind of distributed ledger. Corda, Canton and Hedera agree without a chain of blocks.
  • Distributed describes where ledger copies live, not who controls the records or changes the rules.
  • Shared ledgers can reduce reconciliation, but instant settlement needs cash and securities ready beforehand.

What makes a ledger distributed

Several organizations can keep an agreed record of who owns what, instead of maintaining separate books and reconciling differences later. ISO 22739:2024, published in January 2024, describes shared records synchronized through agreement among network participants. The EU DLT Pilot Regime and MiCA use the same two tests. DLT names the technology that operates these arrangements, rather than any particular record format.

SourceDateDefinition, in its own wordsWhat it adds
ISO 22739:2024DateJanuary 2024Definition, in its own words"ledger that is shared across a set of distributed ledger technology (DLT) nodes and synchronized between the DLT nodes using a consensus mechanism"What it addsA node "stores a complete or partial replica of the ledger records"
EU DLT Pilot Regime and MiCADate30 May 2022, 31 May 2023Definition, in its own words"an information repository that keeps records of transactions" that is "shared across, and synchronised between, a set of DLT network nodes using a consensus mechanism"What it addsThe legal definition in EU law, word for word in both regulations
BIS CPMI analytical frameworkDateFebruary 2017Definition, in its own words"processes and related technologies that enable nodes in a network (or arrangement) to securely propose, validate and record state changes (or updates) to a synchronised ledger"What it addsNo need for a central authority to keep "a single 'golden copy'"
World Bank, FinTech Note No. 1Date2017Definition, in its own words"a novel and fast-evolving approach to recording and sharing data across multiple data stores (or ledgers)"What it adds"simply a distributed database with certain specific properties"
UK Government Chief Scientific AdviserDate19 January 2016Definition, in its own words"essentially an asset database that can be shared across a network of multiple sites, geographies or institutions"What it addsDistributed does not mean "no overall controlling authority or owner"
NIST IR 8202DateOctober 2018Definition, in its own wordsLedgers "implemented in a distributed fashion (i.e., without a central repository) and usually without a central authority"What it addsThey "cannot be considered completely immutable"
Hyperledger Fabric docsDateRead 3 November 2026Definition, in its own words"one logical ledger" with "multiple copies" kept consistent "through a process called consensus"What it addsThe same idea in one platform's words

The formal definitions combine distribution across nodes with synchronization through a consensus mechanism. A network node can hold a complete or partial replica, under ISO 22739:2024. The CPMI framework of February 2017 allows a node to see only transactions involving its own firm. It also removes the necessity for a central party's single "golden copy". Copies need not update simultaneously. The UK report of January 2016 describes updates taking seconds or minutes, rather than every copy changing at once.

Distribution concerns the location of records, while decentralized control concerns decision rights. ISO 22739:2024 separates those concepts. The UK report of January 2016 calls their conflation a misconception. NIST IR 8202, published in October 2018, says blockchains are usually without a central authority, not always. A distributed ledger can have one operator or a small governing group. A permissioned ledger controls who may write. Neither that label nor the public blockchain label alone describes how its governance distributes power.

Distributed ledger, blockchain and replicated database

These distributed ledger categories overlap rather than compete. The World Bank's 2017 note describes DLT as a distributed database with specific properties. The comparison separates record structure from trust assumptions. It also distinguishes keeping copies available from deciding which updates independent parties will accept.

QuestionReplicated databaseDistributed ledgerBlockchain
Who keeps copiesReplicated databaseServers of one organizationDistributed ledgerSeveral parties, each with a full or partial replicaBlockchainSeveral parties or anyone, usually full copies
Faults it plans forReplicated databaseReplicas that crash or stop (Raft)Distributed ledgerParties that may fail or lie, within a stated thresholdBlockchainThe same, with records chained in blocks
How records are orderedReplicated databaseA leader or primaryDistributed ledgerConsensus: BFT ordering, a notary, a sequencer or votingBlockchainMost work, proof-of-stake votes, or BFT ordering
ExamplesReplicated databaseDatabase clusters replicated with RaftDistributed ledgerCorda, Canton, HederaBlockchainBitcoin, Ethereum, Fabric

A blockchain ledger is one kind of distributed ledger, with confirmed blocks linked in an append-only sequence. Other designs do not maintain one chain of blocks. Corda's white paper says it has no chain of blocks. Hedera uses a directed acyclic graph, or DAG, and describes its hashgraph as ACID compliant. Canton provides a "virtual global ledger" with "no physical global state". IOTA's Rebased upgrade on 5 May 2025 replaced its earlier UTXO-based ledger with a Move-based object ledger. Record structure alone does not explain the trust model.

A replicated database usually assumes one organization's replicas fail by stopping, rather than supply conflicting information. Raft covers these non-Byzantine conditions. The Fabric paper of 2018 describes peers that do not fully trust one another. Permissioned designs can use Byzantine fault-tolerant consensus. Fabric documentation notes that full Byzantine protection may be unnecessary within one enterprise. A private blockchain still needs explicit fault assumptions. Wüst and Gervais argued in 2018 that a single writer gains no additional guarantees from blockchain over a regular database.

How distributed ledgers record a transfer

A transfer changes ownership or balances, but each design proves that change differently. The sequence below follows one transfer across five distributed ledgers. It also shows a double spend attempt, when two transactions try to use the same asset. Agreement on order does not mean every node stores identical records.

Figure 1 · Platform documentation, read 3 November 2026Record one transferChoose Bitcoin's chain of blocks, Hedera's hashgraph, Corda, Canton or Fabric, then record a transfer or attempt two spends to inspect four steps.
Ledger
Action
  1. 1SentSign and broadcastAlice signs a transaction that spends an unspent output, and nodes relay it.
  2. 2In a blockMine a blockA miner puts it in a block whose header holds the previous header's hash.
  3. 3AcceptedCheck and extendEvery node checks that the inputs are valid and not already spent, then builds on the chain with the most work.
  4. 4ProbabilisticWait for depthEach new block adds a confirmation. A reversal gets less likely with each one.
  1. 1ConflictTwo spends of one outputAlice signs a second transaction that spends the same output to herself.
  2. 2RejectedOnly one fitsNodes reject any block that spends an output already spent in its chain.
  3. 3ForkCompeting blocksIf two miners pick different versions, nodes follow the chain with more work.
  4. 4ProbabilisticLosing spend droppedThe spend on the shorter chain is reorganized out. This is why merchants wait for confirmations.
  1. 1SentSubmit to a nodeAlice sends a transfer to one of the consensus nodes.
  2. 2In an eventGossip an eventThe node wraps it in an event with two parent hashes and gossips it, so every copy of the hashgraph grows.
  3. 3OrderedVirtual votingEach node works out the votes from its own copy. The consensus timestamp is the median of the times members received it.
  4. 4FinalApply in orderEvery node applies transactions in consensus order, so every copy of the state matches.
  1. 1ConflictTwo transfers of one balanceAlice submits two transfers of the same balance through two different nodes.
  2. 2OrderedTwo timestampsEach transfer gets its own consensus timestamp.
  3. 3AcceptedEarlier applies firstEvery node applies the earlier transfer first, and the balance falls.
  4. 4RejectedLater one failsThe later transfer meets the reduced balance and fails, the same way on every node.
  1. 1ProposedBuild a transactionAlice's node consumes her input state and proposes an output state for Bob.
  2. 2SignedCollect signaturesThe required parties sign. Nodes that are not involved never see it.
  3. 3UniqueNotariseThe notary checks that no other committed transaction consumed the input, then signs.
  4. 4FinalRecordAlice's and Bob's nodes record it. The input state becomes historic.
  1. 1ConflictReuse the inputAlice proposes a second transaction that consumes the same input state for Carol.
  2. 2SignedCarol can signCarol's node cannot see the first transaction, so it may sign.
  3. 3RejectedNotary refusesThe notary has already signed a transaction that consumed this input.
  4. 4BlockedNothing committedWithout the notary's signature the transaction is not committed to the ledger.
  1. 1PreparedPrepare and encryptAlice's participant node prepares the Daml transaction and encrypts it into views for the nodes that need them.
  2. 2OrderedSequenceThe sequencer gives the batch a position and timestamp and delivers views it cannot read.
  3. 3ConfirmedCheck and confirmEach receiving participant checks that the input contracts are still active in its local store, then confirms.
  4. 4FinalVerdictThe mediator collects confirmations and the verdict is sequenced and delivered. There are no forks or rollbacks.
  1. 1ConflictReuse the contractAlice submits a second transaction that consumes the same contract.
  2. 2OrderedOrdered secondThe sequencer places it after the first transaction.
  3. 3RejectedContract archivedThe participants find the contract already archived, so they reject the view.
  4. 4BlockedNo state changeOne signatory's rejection is enough to reject the whole transaction.
  1. 1EndorsedExecute and endorseEndorsing peers run the chaincode against the world state and sign the result, with the versions of the keys they read.
  2. 2OrderedOrderThe ordering service, Raft or SmartBFT, puts the transaction into a block.
  3. 3ValidValidateEach peer checks the endorsement policy and that every key read still has the same version.
  4. 4FinalCommitPeers append the block and update the world state.
  1. 1ConflictTwo spends endorsedTwo transfers are endorsed against the same version of Alice's balance.
  2. 2OrderedBoth orderedThe ordering service puts both into blocks in one order.
  3. 3AcceptedFirst commitsThe first is valid and moves the key to a new version.
  4. 4RejectedSecond failsThe second read an older version, so every peer marks it invalid.
Bitcoin, chain of blocks
What links records
Each block header holds the hash of the previous header and a Merkle root of its transactions
Who keeps what
Every full node checks every block. Pruned nodes delete old ones
What stops a double spend
Valid, unspent inputs only, and the chain with the most work
When it is final
Probabilistic. Six confirmations take about an hour
Hedera, hashgraph
What links records
Each event holds two parent hashes: its own last event and the last event of the member it synced with
Who keeps what
Each consensus node holds the state and the recent hashgraph. Mirror nodes serve the history
What stops a double spend
One consensus order, then balance checks in that order
When it is final
Once ordered and applied, "there is no chance of reversal"
Corda
What links records
Each input state points to an output of an earlier transaction
Who keeps what
Each node holds only the facts it is party to
What stops a double spend
The notary's uniqueness check on input states
When it is final
"notarisation is the point of finality"
Canton
What links records
A Daml transaction tree is split into encrypted views, and the sequencer timestamps each batch
Who keeps what
Each participant node holds only the contracts of its hosted parties
What stops a double spend
Each confirming participant checks that inputs are still active, in sequencer order, within a single synchronizer
When it is final
"Once the verdict is sequenced and delivered, the transaction result is final."
Hyperledger Fabric
What links records
Blocks are hash-linked, and each world-state key carries a version
Who keeps what
Each peer holds the ledger of its channels. Private data reaches only collection members
What stops a double spend
The read-set version check at validation
When it is final
"any block validated by the peer is guaranteed to be final and correct"
Who stores this transferSix-node exampleAlice's nodeStores itReceiving nodeApplies itAlice's nodeStores itAlice's participantStores itPeer, org AStores itBob's nodeStores itConsensus nodeApplies itBob's nodeStores itBob's participantStores itPeer, org BStores itNode CStores itConsensus nodeApplies itNotaryChecks the inputsSequencerOrders, cannot readPeer, org CStores itNode DStores itConsensus nodeApplies itNode DNever sees itMediatorCollects confirmationsOrdering nodeOrders and keeps the blockA minerStores itConsensus nodeApplies itNode ENever sees itParticipant ENever sees itPeer, other channelNever sees itNode FStores itMirror nodeKeeps the historyNode FNever sees itParticipant FNever sees itPeer, other channelNever sees it
Every full node receives and checks the new block, so all six copies change.
Every consensus node applies the transfer to its copy of the state. Mirror nodes keep the history.
Only the parties store the transaction. The notary checks that the input was not used before.
Only the participants hosting Alice and Bob store the contracts. The sequencer orders views it cannot read.
Peers and ordering nodes on the channel store the block. Peers on other channels never receive it.

Bitcoin nodes reject a double spend and follow the chain with the most work. Finality remains probabilistic rather than an immediate guarantee.

Hedera nodes apply transactions by consensus timestamp. A later conflicting spend meets the reduced balance, and ordered, applied transactions have no chance of reversal.

Corda parties consume input states and create output states. A notary checks uniqueness before commitment, without broadcasting every transaction to every participant.

Canton's sequencer orders messages, while participants check active contracts. Once the verdict is sequenced and delivered, the transaction result is final.

Fabric uses endorsement, ordering and validation. A conflicting read-set version fails validation, and a block validated by a peer is final.

These steps summarize platform documentation read on 3 November 2026, simplified by the page team to one transfer.

Bitcoin block headers link to previous headers through cryptographic hashes and contain a Merkle root of transactions. Hedera events carry two parent hashes. Each transaction's consensus timestamp is the median of members' reported first-receipt times. Corda transactions consume existing states and produce new ones. Canton participants check transaction views against their own active contracts, with agreements expressed in Daml contracts. Hyperledger Fabric combines a world state with a blockchain. Its execute-order-validate process compares read-set versions with current key versions after ordering.

Double spend prevention depends on these checks, not just record links. Bitcoin nodes accept valid, unspent transactions in the most-work chain. Hedera applies its consensus order, while Corda's notary attests that inputs remain unused. Canton checks active contracts to prevent reuse within a single synchronizer. Fabric rejects a transaction whose read-set versions no longer match. IOTA locks owned object versions when validators sign. Finality also differs, from Bitcoin's probabilistic chain to Canton's final verdict without forks, reorganizations or rollbacks.

Shared records and reconciliation

A broker, investment manager and custodian can disagree about the same bond trade. The model compares three databases with two distributed ledger designs. One requires counterparty confirmation, while the other accepts the submitter's signature alone. This distinction determines whether a quantity error is challenged before commitment or simply becomes a shared mistake.

Figure 2 · Page team model, three firms and one tradeFind the record mismatchChoose a quantity typo, lost confirmation or racing updates, then compare designs to see each firm's record, the mismatch, detection and correction.
Error
Design
Agreed order: buy 10,000 units at 99.50Agreed trade: booked by the broker and the investment manager, to be held by the custodianTwo updates: price 99.50, then an amendment to 99.40
BrokerDiffers1,000 units at 99.50
Investment managerMatches order10,000 units at 99.50
CustodianMatches order10,000 units at 99.50
What the firms seeRecords disagree
Wrong records
1 of 3
When it is caught
At affirmation, if the manager compares the broker's confirmation, or later as a settlement fail
What fixes it
An exception handled between the firms, then a corrected confirmation
BrokerRejected1,000 units proposed
Investment managerRejectedDeclined, expects 10,000
CustodianNot affectedNo change
What the firms seeStopped before recording
Wrong records
0 of 3
When it is caught
Before anything is recorded, when the manager declines to confirm
What fixes it
The broker resubmits 10,000 units and both parties confirm
BrokerShared1,000 units at 99.50
Investment managerShared1,000 units at 99.50
CustodianShared1,000 units at 99.50
What the firms seeShared mistake
Wrong records
3 of 3
When it is caught
Only when someone compares the ledger with the order kept outside it
What fixes it
A new correcting transaction. The ledger cannot tell whether input data is true
BrokerMatchesTrade booked
Investment managerMatchesTrade booked
CustodianMissingNo record
What the firms seeRecords disagree
Wrong records
1 of 3
When it is caught
At reconciliation, or when settlement fails
What fixes it
Resend the confirmation and reconcile the custodian's books
BrokerMatchesTrade committed
Investment managerMatchesTrade committed
CustodianMatchesTrade committed
What the firms seeNo mismatch
Wrong records
0 of 3
When it is caught
Nothing to catch: the transaction is committed for every party or not at all
What fixes it
None needed
BrokerMatchesTrade committed
Investment managerMatchesTrade committed
CustodianMatchesTrade committed
What the firms seeNo mismatch
Wrong records
0 of 3
When it is caught
Nothing to catch: the transaction is committed for every party or not at all
What fixes it
None needed
BrokerMatchesAmended price 99.40
Investment managerDiffersOriginal price 99.50
CustodianMatchesAmended price 99.40
What the firms seeRecords disagree
Wrong records
1 of 3
When it is caught
At reconciliation, after each firm applied the updates in its own order
What fixes it
Agree which update came last, then correct the other books
BrokerMatchesAmended price 99.40
Investment managerMatchesAmended price 99.40
CustodianMatchesAmended price 99.40
What the firms seeNo mismatch
Wrong records
0 of 3
When it is caught
At ordering: both updates get one agreed order, and a stale one is rejected
What fixes it
None needed
BrokerMatchesAmended price 99.40
Investment managerMatchesAmended price 99.40
CustodianMatchesAmended price 99.40
What the firms seeNo mismatch
Wrong records
0 of 3
When it is caught
At ordering: both updates get one agreed order, and a stale one is rejected
What fixes it
None needed
US post-trade figures for context, not DLT results
Affirmed on time, end of January 202473%US trades, DTCC
Affirmed on time, 29 May 202494.55%First T+1 settlement day
CNS fail rate, July 20242.12%SIFMA, ICI and DTCC report

Counterparty confirmation can reject a wrong quantity before recording. When only the submitter signs, the shared ledger copies the same mistake to participating nodes.

In the shared model, a lost confirmation cannot leave conflicting committed records. The trade commits for every party or not at all.

Two racing updates receive one agreed order on the shared ledger. Separate databases can apply them differently until the firms reconcile their records.

This page team model uses three firms and one trade, not measured error rates. DTCC figures provide real context.

The World Bank's 2017 "shared truth" argument suggests removing counterparty reconciliation. CPMI's February 2017 framework says DLT may reduce discrepancies. DTCC's US timely affirmation rate rose from 73% at the end of January 2024 to 94.55% on 29 May 2024. The September 2024 report says nearly 95%. Fail rates hovered around 2% in May and July 2024. These are not DLT results. The ECB's April 2016 paper rejects redundant storage as the back-office cost driver. NIST's October 2018 report cautions that a ledger cannot verify real-world input truth.

Settlement on a shared ledger

A distributed ledger does not set the settlement timetable. Hong Kong's fourth digital green bond offering, dated 29 September 2026, settled through the CMU on a T+1 basis. The model separates a shorter waiting period from exchanging cash and securities together.

Figure 3 · Page team model from settlement rules and BIS papersCompare three settlement pathsChoose T+2, T+1 or atomic delivery versus payment, then follow holdings, default exposure, netting and pre-funding from trade through settlement.
Settlement path
1 of 4
Trade date (T). Terms agreed. Nothing moves yet.
Affirmation. The firms match the trade details.
Clearing and netting (T+1). Obligations are netted across many trades.
Settlement (T+2). Bond and cash move through delivery versus payment.
Trade date (T). Terms agreed. Nothing moves yet.
Affirmation (T). The firms match the details on trade date.
Clearing and netting (T). Obligations are netted the same evening.
Settlement (T+1). Bond and cash move through delivery versus payment.
Before the trade. Both sides place assets on the ledger first.
Trade (T). A smart contract holds the agreed terms.
Atomic swap. Both legs execute together, or neither does.
Settled (T). Settlement happens when the trade is executed.
BuyerExposedCashOpen to the other side's default until settlement
SellerExposedBondOpen to the other side's default until settlement
BuyerExposedCashOpen to the other side's default until settlement
SellerExposedBondOpen to the other side's default until settlement
BuyerExposedCash, netted obligationOpen to the other side's default until settlement
SellerExposedBond, netted obligationOpen to the other side's default until settlement
BuyerSettledBondDelivery versus payment complete
SellerSettledCashDelivery versus payment complete
BuyerExposedCashOpen to the other side's default until settlement
SellerExposedBondOpen to the other side's default until settlement
BuyerExposedCashOpen to the other side's default until settlement
SellerExposedBondOpen to the other side's default until settlement
BuyerExposedCash, netted obligationOpen to the other side's default until settlement
SellerExposedBond, netted obligationOpen to the other side's default until settlement
BuyerSettledBondDelivery versus payment complete
SellerSettledCashDelivery versus payment complete
BuyerPre-fundedCash token, locked for the tradeLocked on the ledger for this trade
SellerPre-fundedBond token, locked for the tradeLocked on the ledger for this trade
BuyerPre-fundedCash tokenLocked on the ledger for this trade
SellerPre-fundedBond tokenLocked on the ledger for this trade
BuyerExecutingSwappingBoth legs move together, or neither does
SellerExecutingSwappingBoth legs move together, or neither does
BuyerSettledBond tokenDelivery versus payment complete
SellerSettledCash tokenDelivery versus payment complete
Time from trade to settlement2 business daysT+2
NettingYesObligations offset across many trades
Pre-fundingNoAssets needed only at settlement
Where it applies
US until 27 May 2024. EU, UK and Switzerland until 10 October 2027
NSCC Clearing Fund
US$12.8 billion under T+2, US$9.8 billion under T+1. Down 23%, SIFMA, ICI and DTCC report, 12 September 2024
Time from trade to settlement1 business dayT+1
NettingYesObligations offset across many trades
Pre-fundingNoAssets needed only at settlement
Where it applies
US since 28 May 2024. EU, UK, Switzerland and Liechtenstein from 11 October 2027
NSCC Clearing Fund
US$12.8 billion under T+2, US$9.8 billion under T+1. Down 23%, SIFMA, ICI and DTCC report, 12 September 2024
Time from trade to settlementSame momentAtomic DvP on a shared ledger
NettingNoEach trade settles gross
Pre-fundingYesCash and bond placed on the ledger first
Where it applies
Tokenised MAS Bills in wholesale CBDC, SDX bonds against SNB wCBDC, Pontes across linked ledgers from 21 September 2026
NSCC Clearing Fund
Not part of this path. Each trade settles gross, so each side pre-funds instead of netting.

The model leaves two business days between trade and settlement. Netting reduces obligations, but counterparties remain exposed during the waiting period.

The model shortens that interval to one business day while retaining netting. A shorter cycle does not require a distributed ledger.

Cash and securities move together or neither moves. In this model, removing netting means the buyer and seller must have assets ready beforehand.

This page team model uses the cited settlement rules and BIS papers, simplified to one bond trade.

The US standard is T+1 from 28 May 2024. The EU, UK, Switzerland and Liechtenstein move to T+1 on 11 October 2027. The SIFMA, ICI and DTCC report of 12 September 2024 found that the NSCC Clearing Fund fell 23%, from US$12.8 billion to US$9.8 billion. CPSS's September 1992 delivery versus payment models address principal risk: payment or delivery without its counterpart. Model 1 settles both legs gross, simultaneously. Model 2 settles securities gross and funds net, while Model 3 nets both. Net obligations settle at cycle end.

Atomic settlement executes every leg or none. BIS and CPMI's October 2024 paper identifies higher liquidity needs without netting. Parties must pre-fund cash and position securities, as the New York Fed explained in November 2022. CPMI's February 2017 framework defines finality as a legal moment, potentially less clear for DLT. Pontes, launched on 21 September 2026, connects DLT trades to central bank money. Other examples through 3 November 2026 include SDX's Project Helvetia securities, tokenized MAS Bills from September 2026 and EnsembleTX from 2 November 2026.

Distributed ledgers in financial markets in 2026

The November 2026 table separates repo, collateral, securities issuance and payments, rather than treating them as one market. Clearstream and SDX illustrate tokenized securities infrastructure. Project Agorá settled real-value transactions in July 2026 and was still ongoing in November 2026. Published volumes use different measures and should not be added together.

ArrangementWhat it recordsLedgerLatest public figure
Broadridge DLRWhat it recordsRepo transactionsLedgerCanton Network, per canton.networkLatest public figure$7.5 trillion in September 2026, $359 billion per business day
DTCC Tokenization ServiceWhat it recordsTokenized DTC-held assetsLedgerCanton Network and DTCC's private Besu networkLatest public figureLive on 28 October 2026, 20 participants signed
J.P. Morgan KinexysWhat it recordsPayments, deposits and digital assetsLedger"private-permissioned asset platform"Latest public figureOver $7 billion average daily volume (2025 data)
HQLAxWhat it recordsCollateral ownership at a custodianLedgerCorda Enterprise, move to Canton plannedLatest public figureSEC staff no-action letter, 4 May 2026
Clearstream D7 DLTWhat it recordsTokenized securitiesLedgerNot namedLatest public figureLaunched 4 November 2025. EIB commercial paper, EUR 77.5 million, 29 June 2026
SIX Digital ExchangeWhat it recordsDigital securities, a regulated CSDLedgerNot named on its pageLatest public figureOver CHF 2 billion issued (read 3 November 2026)
Euroclear D-FMIWhat it recordsDigital Native NotesLedger"based on Distributed Ledger Technology (DLT)"Latest public figureDvP settlement in EUR or USD (read 3 November 2026)
Fnality £FnPSWhat it recordsWholesale sterling paymentsLedgerDLT with a Bank of England omnibus accountLatest public figureFirst live transactions, 14 December 2023
BIS Project AgoráWhat it recordsCross-border payments, tokenised reserves and depositsLedgerPrototype platformLatest public figureReal-value transactions in July 2026, ongoing
HKMA EnsembleTXWhat it recordsTokenised deposits and digital assetsLedgerNot namedLatest public figure24/7 tokenised central bank money from 2 November 2026
ECB PontesWhat it recordsCentral bank money for DLT tradesLedgerLinks DLT platforms to TARGET servicesLatest public figureLive on 21 September 2026 with four DLT operators
Tokenized US TreasuriesWhat it recordsFunds and notesLedgerPublic chainsLatest public figure$14.21 billion in 119 assets (rwa.xyz, 3 November 2026)

Broadridge DLR reported $7.5 trillion for September 2026. Kinexys reported over $7 billion in average daily volume in 2025. DTCC's tokenization service went live on 28 October 2026 on Canton and its private Besu network. HQLAx announced a planned Corda-to-Canton migration on 21 April 2026, subject to regulatory approval. Pontes connects DLT platforms to central bank money from September 2026. Yet IOSCO's November 2025 report found 91% of surveyed regulators reported nil or very limited tokenization use cases. These distributed ledger figures measure different activities.

Laws written for distributed ledgers

Legal recognition concerns rights, operators and settlement arrangements, not merely the software storing a record. The November 2026 table separates legislation, supervised testing and staff no-action relief. These are different legal instruments. A distributed ledger's technical finality does not, by itself, establish which rights a tokenized security represents.

JurisdictionFrameworkDatesWhat it allows
European UnionFrameworkDLT Pilot Regime, Regulation (EU) 2022/858DatesApplies from 23 March 2023What it allowsDLT trading and settlement under caps of EUR 6 billion per infrastructure, with a transition plan at EUR 9 billion. Six infrastructures authorised
SwitzerlandFrameworkDLT ActDates1 February and 1 August 2021What it allowsLedger-based securities and DLT trading facilities. BX Digital licensed on 18 March 2025
GermanyFrameworkElectronic Securities Act (eWpG)Dates10 June 2021What it allowsCrypto securities registers on forgery-proof, time-ordered recording systems, with BaFin authorisation
United KingdomFrameworkDigital Securities SandboxDates8 January 2024 to 8 January 2029What it allows17 firms through Gate 1 and 3 through Gate 2 (2 November 2026)
United StatesFrameworkSEC staff letter to DTC, Nasdaq approvalDates11 December 2025, 18 March 2026What it allowsTokenized entitlements for three years from launch, and trading of securities in tokenized form
SingaporeFrameworkProject Guardian, tokenised MAS BillsDatesBills issued from 23 September 2026What it allowsIndustry frameworks for fixed income and funds. Bills settled in wholesale CBDC
Hong KongFrameworkProject EnsembleDatesPilot through 2026What it allowsReal-value transactions with tokenised deposits. 24/7 tokenised central bank money from 2 November 2026

The EU DLT Pilot Regime applies from 23 March 2023, with a EUR 6 billion cap and EUR 9 billion transition trigger. ESMA's list names six authorised market infrastructures, approved between October 2024 and November 2025. ESMA suggested on 25 June 2025 making the regime permanent. The Commission proposed on 4 December 2025 to broaden activity and make the regime more flexible. The Council agreed its negotiating position on 9 October 2026. Those proposals are distinct from the regime's applicable limits.

Switzerland's DLT Act enables ledger-based securities from February 2021, and FINMA licensed BX Digital on 18 March 2025. Germany's eWpG, effective 10 June 2021, requires protected, time-ordered registers. The UK Digital Securities Sandbox dashboard dated 2 November 2026 lists 17 Gate 1 firms and three Gate 2 firms. US measures include DTC's 11 December 2025 staff no-action letter and the SEC's 18 March 2026 approval of Nasdaq's rule change. These approaches define distributed ledger functions and regulated activities, rather than endorse a software brand.

Limits of distributed ledgers

CPMI's February 2017 framework says ledger replication can reduce single-point failure risk but adds attack entry points. The Federal Reserve's 2016 study notes lost keys can block access. IOSCO's November 2025 report found 91% of surveyed regulators reported nil or very limited use. ESMA noted initially limited uptake in June 2025.

Legal certainty and finality

CPMI's February 2017 framework warns that asset ownership and parties' obligations may be unclear on a distributed ledger. Settlement finality requires a legally irrevocable, unconditional transfer, not just technical agreement.

Liquidity and pre-funding

BIS and CPMI explained in October 2024 that instant gross settlement needs more liquidity than netting. Atomic delivery versus payment on a shared ledger addresses principal risk, not the funding requirement.

Fragmentation and interoperability

The ECB warned on 26 August 2026 that incompatible ledger platforms can deepen fragmentation. IOSCO identified limited cross-blockchain interoperability in November 2025. Technical connections also need legal compatibility.

Privacy and confidentiality

CPMI's February 2017 framework balances traceability against confidentiality. The EDPB's July 2026 guidance generally advises against storing personal data on a blockchain, including ledger records with restricted access.

Governance and central roles

Notaries, ordering services, Canton's sequencer and mediator, and governing councils remain part of ledger arrangements. Permissioned networks still need rules for participation, upgrades and operator accountability.

Adoption and project failures

ASX's November 2022 pre-tax charge was A$245 to 255 million. The Accenture review found ASX remained the central source of truth and that using the ledger mainly for resiliency added undue complexity.

Building on a distributed ledger

We frame distributed ledger engineering around records and responsibilities. PixelPlex is a software engineering company founded in 2007, building in blockchain since 2013. On Canton, we are a validator operator and run our own production infrastructure, and our Canton development services cover participant nodes and Daml applications. The decisions below apply before choosing a permissioned blockchain or implementing transaction logic.

PixelPlex case icon

Our Flact case is a permissioned blockchain built on Hyperledger Fabric for Rinne Technologies: a DLT network for document storage with a multi-tier access restriction system. Our 5N ID case for FiveNorth on Canton (2025) describes two custom Daml smart contracts that automate verification and consent on the blockchain.

  1. Records and access

    We define who submits, confirms and sees each ledger record, including which nodes need full replicas and which retain only transactions involving their hosted parties.
  2. Trust and faults

    We specify the faults consensus must tolerate, the role of each ledger operator and how signatures establish responsibility between organizations that do not fully trust one another.
  3. Settlement and finality

    We distinguish legal completion from technical agreement for tokenized assets, define settlement assets and assess pre-funding needs. Stablecoin mechanics are separate from a token's legal treatment.
  4. Integration and legal terms

    We map existing database records to ledger transaction inputs and outputs, then identify where smart contracts, external systems and legal agreements must describe the same business event.
  5. Operations and governance

    We document responsibilities for keys, upgrades and rule changes, alongside the recovery process. Governance covers both the running software and decisions made outside the distributed ledger.

Discuss the requirements behind your shared records

We can discuss your recordkeeping process, participant roles and settlement requirements before narrowing the engineering scope.

Discuss your requirements

Distributed ledger FAQ

What is a distributed ledger in simple terms?

A distributed ledger is a record of transactions that several nodes keep in copies and keep in step through a consensus mechanism. No single database is the only official version of the ledger. Some nodes hold only their own ledger records. DLT is the technology that runs this shared recordkeeping.

How does a distributed ledger differ from blockchain?

A blockchain is one kind of distributed ledger, in which confirmed blocks are linked by hashes in one append-only chain. Corda, Canton's virtual global ledger and Hedera's hashgraph reach agreement without building one chain of blocks. Each design still needs a way to stop a double spend. Visibility, control and finality follow the network's rules and transaction process.

Is a distributed ledger just a distributed database?

Not quite. A replicated database usually assumes copies run by one organization that fail by stopping. A distributed ledger is usually kept by parties who do not fully trust each other. Consensus and checkable ledger records support agreement between those parties. The World Bank's 2017 note also explains DLT in database terms, so the terminology is not absolute.

Does every node hold the whole ledger?

Not always. ISO 22739:2024 lets a node hold a complete or partial replica of the ledger. In Corda, peers see a subset of facts, while Canton participants store contracts involving their hosted parties. Fabric channels also limit information sharing. A node's ledger view depends on the arrangement, so shared records need not disclose every firm's transactions to all participants.

How is DLT used in finance in 2026?

In 2026, distributed ledgers run repo trades, collateral transfers, tokenized securities and payments. Broadridge reported $7.5 trillion in repo activity in September 2026. DTCC's tokenization service went live on 28 October 2026 for representations of DTC-held assets. Project Agorá settled real-value cross-border payment transactions in July 2026. These examples describe different activities, not one interchangeable measure of distributed ledger adoption.

Which laws and programs cover distributed ledgers?

Laws written for distributed ledgers include the EU DLT Pilot Regime, Switzerland's DLT Act, Germany's eWpG and the UK's Digital Securities Sandbox. Their scope differs, so technical labels do not establish regulatory treatment. In the US, DTC's pilot relies on SEC staff no-action relief dated 11 December 2025. Since 30 June 2026, UK sandbox firms can apply to use specific stablecoins as a settlement asset.

What are the main risks of distributed ledger technology?

The main risks of distributed ledger technology are unclear legal rights and settlement finality, liquidity needs, fragmentation between platforms, privacy and governance. More nodes can also create additional attack entry points. A ledger cannot tell whether input data is true, regardless of ledger immutability. Shared ledger records still depend on sound inputs, controlled access and clear responsibilities for the people operating the arrangement.
Sources (34)

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