Scaling Executive Summary

  • Scaling Landscape

  • How will this be applicable to Tari?

  • Scaling Context for Tari

  • Scaling Alternatives for Tari

  • Observations

See Layer 2 Scaling Survey for the full report.

© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)

Scaling Landscape

  • Layer 2 Scaling can be roughly grouped into these groups:
    • Off-chain matching engines
    • Off-chain processing nodes
    • Off-chain payment channels
    • Two Way Pegged (2WP) secondary chains
    • Tiered block chains
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)

L2ScalingLandscape w:900px

© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)
  • Other scaling technologies investigated, mainly directly on the primary block chain
    • DAG derivative protocols
    • Transaction data compression
    • Scriptless scripts

Non_L2S w:700px

© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)

How will this be Applicable to Tari?

  • Tari is a high-throughput protocol -> tens of thousands of transactions per second imagined
  • This will be impossible to do with primary block chain scaling solutions alone
  • Example use cases:
    • EventCorp, the stadium owner
      • Selling: Do not let the servers crash!
      • Redeeming:
        • 85,000 seats, 72 queues tickets on match days
        • scanning 500 tickets in 4 minutes, that is ~2 spectators allowed access per second per queue
    • Steve, the collectable card game entrepreneur
      • Steve created his own digital collectible
      • Steve developed a website with the Tari API to display his digital cards, update their status in real time, and facilitate trading and interacting
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)

Scaling Context for Tari

  • Scaling technologies worth investigating further
    • TumbleBit, as an off-chain matching engine (L2S)
    • Federated Nodes/Masternodes as off-chain processing nodes (L2S)
    • Scriptless Scripts & Schnorr Signature Aggregation, as Layer 1 scaling
    • SPECTRE, PHANTOM, as DAG derivative protocol alternative to a traditional block chain, also Layer 1 scaling
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)


© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)

#1 Why TumbleBit?


  • Notes:
    • The most important Bitcoin functionality used here are hashing conditions, signing conditions, conditional execution, 2-of-2 multi signatures and timelocking
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)
  • TumbleBit has many excellent properties as trustless matching engine
  • TumbleBit can perform off-chain payments in batch mode
  • Commercial implementation NTumbleBit far advanced, backed by Boston University that provided proof-of-concept and reference implementation alongside white paper
  • Anonymity & bad acting prevention provided by RSA-Puzzle-Solver Protocol & Puzzle-Promise Protocol, making use of RSA crypto blinding properties
  • TumbleBit also supports anonymizing through Tor
  • Notes:
    • TumbleBit is tailor made for Bitcoin; not all features that make it work currently part of Mimblewimble
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)

#2 Why Federated Nodes/Masternodes?


© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)
  • A federated node is a special case of a masternode, with emphasis on the federated trust model.
  • Federated Nodes provides a protocol and network of nodes for creating smart contract applications using a customized virtual machine or other mechanism and linked to the primary block chain.
  • All smart contracts and their state updates are executed and maintained off-chain in the federated nodes.
  • The Federated Node software stack model lends itself for high volume processing.
  • Federated Nodes does not have to use embedded consensus (although Counterparty does); improved consensus models like Federated Byzantine Agreement (FBA) can be implemented.
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)

#3 Why Scriptless Scripts & Schnorr Sig. Aggregation?


© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)
  • Data savings: Signature aggregation using an appropriate Schnorr-based multi-signature scheme (e.g. MuSig) provides data compression on the block chain
  • Privacy: Nothing about the Scriptless Script smart contract, other than the settlement transaction,  is ever recorded on the block chain. No one will ever know that an underlying smart contract was executed.
  • Multiplicity: Multiple digital assets can be transferred between two parties in a single settlement transaction.
  • Implicit scalability: Scalability on the block chain is achieved by virtue of compressing multiple transactions into a single settlement transaction. Transactions are only broadcasted to the block chain once all preconditions are met.
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)
  • Signature aggregation properties sought here are:
    • Must be provably secure in the plain public-key model;
    • Must satisfy the normal Schnorr equation, whereby the resulting signature can be written as a function of a combination of the public keys;
    • Must allow for Interactive Aggregate Signatures (IAS) where the signers are required to cooperate;
    • Must allow for Non-interactive Aggregate Signatures (NAS) where the aggregation can be done by anyone;
    • Must allow each signer to sign the same message;
    • Must allow each signer to sign their own message.


  1. Plain public-key model: allows for signature aggregation, mathematically combining several
    signatures into a single signature, without having to prove Knowledge of Secret Keys (KOSK).
  2. KOSK requires that users prove knowledge (or possession) of the secret key during public key
    registration with a certification authority.
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)
  • Possible software stack:

    • The Mimblewimble Scriptless Scripts logic could be implemented by Federated Nodes with FBA on layer 2
    • The MuSig Schnorr-based multi-signature scheme with key aggregation can be used
    • Secrets revealed by virtue of the MuSig Schnorr signatures can instantiate normal smart contracts inside the Federated Nodes, with intermediate state updates confirmed by FBA
  • Challenges:

    • No space allowed to embed data other than Tx inputs and outputs in the form of coin amounts
    • Smart contract state updates can't be written back to the block chain after the event
    • Immediate cut-through (& pruning) may delete transactions intended to be persistent
    • HTLC not supported
    • Standard Mimblewimble Tx does not support signaling a Federated Node
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)



© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)
  • DAG derivative protocols SPECTRE and PHANTOM offer an alternative to a block chain, i.e. block DAG
  • Strengths:
    • Layer 1 scaling: Increased transaction throughput on the main block chain

    • Fairness: Better payoff for weak miners

    • Decentralization mitigation: Weaker miners also get profits

    • Transaction confirmation times: Confirmation times of several seconds (SPECTRE)

    • Smart contracts: Support smart contracts (PHANTOM)

© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)


  • High throughput and fast confirmation times [GOOD]
  • Weak liveness for conflicting transactions [BAD]
  • DAG structure represents an abstract vote regarding the order between each pair of blocks [BAD]


  • Confirmation times are mush slower than those in SPECTRE [BAD]
  • Strong liveness for conflicting transactions [GOOD]
  • Linear ordering over the blocks of the DAG and can support consensus regarding any general computation (smart contracts) [GOOD]
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)

SPECTRE and PHANTOM can be combined


  1. Strong liveness: all conflicts decided in finite time
  2. Weak liveness: no guarantee a resolution can be reached for conflicting transactions published soon one after the other
© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)


  • Not all protocols presented here have production or even reference implementations

  • Careful consideration about each aspect of these technologies and their applicability to the Tari protocol is required

  • Some protocols are tailor made for Bitcoin; features that make it work currently not part of Mimblewimble

© Tari Labs, 2018-2021. (License : CC BY-NC-SA 4.0)