Development Fund Proposal: ISS-Based BFT

Development Fund Proposal

Enhanced Global Synchronizer Scale via Parallel Ordering Consensus

Field	Value
Author	Wayne Collier, Digital Asset
Status	Approved
Created	2026-02-15
Approved	2026-03-23

---

Abstract

This proposal requests funding to develop, release, and deploy a production-ready BFT (Byzantine Fault Tolerant) ordering service for the Canton protocol. This new ordering service will replace the current single-leader CometBFT ordering path on the Global Synchronizer with a parallel, multi-leader consensus architecture, enabling significantly higher transaction throughput and fault tolerance while preserving Canton's strong privacy, safety, and liveness guarantees.

The core innovation is the separation of request dissemination from consensus ordering. By implementing an availability module inspired by Narwhal that reliably disseminates transaction batches before consensus, and by employing ISS (Insanely Scalable State machine) protocol to run multiple parallel consensus instances, the ordering service removes the throughput bottleneck inherent in single-leader designs. Validators will concurrently produce blocks across independent consensus segments, with a deterministic post-ordering timestamp assignment ensuring globally consistent, strictly increasing timestamps for all transactions.

The implementation also includes a comprehensive governance model integrated with Canton's existing topology management system, a pruning and persistence strategy that satisfies international data-privacy and retention requirements and supports crash recovery, and a fair resource allocation framework that protects correct nodes from abuse by Byzantine peers.

Digital Asset expects that when implemented on the Global Synchronizer, this work will remove the sequencer backend as a bottleneck towards achieving 2500 TPS or more via the Global Synchronizer. The reader should refer to a separate grant proposal on scaling the global synchronizer for specific scaling targets.

Motivation

As the Canton Network grows in participants, transaction volume, and application diversity, the Global Synchronizer's transaction throughput becomes the primary constraint on network scale. The current CometBFT ordering architecture processes transactions through a single-leader consensus path, creating a fundamental bottleneck that cannot be overcome by incremental optimization alone.

This throughput limitation creates several critical issues for the Canton Network's continued growth:

Single-Leader Bottleneck: In traditional BFT consensus (e.g., PBFT, HotStuff), a single leader proposes blocks containing full transaction data. Both the leader's outbound bandwidth and the consensus protocol's communication overhead become throughput ceilings. As transaction volume grows, this ceiling becomes the binding constraint on network utility.

Consensus-Bound Dissemination: When transaction data is disseminated as part of the consensus protocol, the size of proposals directly impacts consensus latency. Larger blocks mean longer rounds; longer rounds mean lower throughput. This coupling forces an undesirable trade-off between block size and confirmation time.

Underutilized Ordering Resources: In a single-leader design, only one Super Validator is actively producing blocks at any given time. The remaining Super Validators participate in voting but their compute, bandwidth, and storage resources are substantially underutilized. As the validator set grows with network adoption, this inefficiency becomes more pronounced.

Governance Fragmentation: Without tight integration between BFT ordering governance and Canton's existing topology management, operational changes such as adding or removing sequencers require out-of-band coordination, manual intervention, and risk of misconfiguration. This creates operational fragility that is inconsistent with the Canton Network's maturity trajectory.

Specification

1. Objective

We propose a new, production-ready BFT ordering service in the Canton open-source repository, capable of deployment on the Global Synchronizer. The technical scope includes four major subsystems:

Mempool & Availability Module: A request dissemination layer that reliably broadcasts transaction batches to all correct ordering nodes before consensus begins. This removes bulk data transfer from the consensus critical path, enabling consensus to operate on lightweight batch references rather than full transaction payloads. The module provides storage fairness guarantees, per-node rate limiting, and garbage collection based on sequencing request expiration times.

Parallel Ordering via ISS: An implementation of segment-based parallel consensus where multiple Super Validators concurrently lead independent consensus instances within each epoch. The log is divided into epochs, each with a configurable number of slots distributed across segments. This architecture utilizes network resources more efficiently than single-leader protocols and provides ordering fairness guarantees.

BFT Timestamping: A median-based, post-order shifting timestamp assignment that produces globally strictly increasing timestamps across all parallel consensus instances. Each validator computes BFT time using 2f+1 signed protocol messages from its sequential blocks, with deterministic adjustment to maintain monotonicity across the interleaved global output.

Governance, Pruning & Persistence: Full integration with Canton's existing topology management system for BFT ordering governance, including online reconfiguration at epoch boundaries, sequencer onboarding, and dynamic parameter updates. A multi-phase pruning strategy that evolves from manual operator-driven pruning to fully automatic, decentralized pruning tolerant of up to f Byzantine nodes.

Fault Tolerance: With classic BFT consensus underpinning ordering and the whole-node fault assumption, a sequencer set of size N can tolerate at most f faults where 3f < N.

2. Implementation Mechanics

2.1 Mempool and Availability

The mempool module serves as the entry point from the sequencer driver, queuing ordering requests and producing batches on demand for the availability module. In the initial design, no gossip protocol is required to initiate ordering; the availability module handles all peer dissemination.
¹
The availability module disseminates request batches to all correct BFT ordering nodes, collecting proofs of availability (PoAs) before batches enter consensus. This follows the key insight from the Narwhal paper: removing dissemination from the critical path of consensus yields significantly higher throughput even with non-DAG consensus algorithms. The design proposes a home-grown approach—using Narwhal-style PoAs without the full DAG and round structure—which is simpler to implement while still providing reliable dissemination. Storage fairness is enforced through per-ordering-node limits on unordered, unexpired proofs of availability.

Each batch includes a deterministic expiration time derived from the contained requests' max_sequencing_time fields. Correct nodes reject expired batch proposals and ordering blocks containing expired batch references, enabling safe garbage collection at epoch boundaries.

2.2 Parallel Ordering via ISS

The ordering layer implements ISS, where the log is divided into epochs, each with k slots distributed across s segments (typically matching the number of active ordering validator nodes). Within each segment, a designated Super Validator leads sequential consensus instances. Across segments, consensus proceeds in parallel, yielding throughput proportional to the number of validators rather than being bottlenecked by a single leader.

The consensus module operates on lightweight batch references provided by the availability module, producing ordering blocks. Once ordered, the output module replaces batch references with actual request data and delivers output blocks to the sequencer logic. Deduplication of requests is handled post-ordering by the output module or by Canton's existing sequencer layer.

2.3 BFT Timestamping

Timestamps are assigned using a median-based approach with post-order shifting. Each validator produces BFT time in isolation for the consensus instances it leads. For each block, the validator includes a canonical set of 2f+1 signed protocol messages from its most recently completed block. All nodes deterministically compute the block timestamp as the median of the timestamps in this canonical set, shifted upward if necessary to maintain strict monotonicity across the globally interleaved output.

Per-transaction timestamps are assigned by incrementing from the block timestamp at microsecond granularity, supporting up to 1,000 transactions per block with a 1-millisecond minimum block time granularity. This approach ensures that all validators deterministically agree on every timestamp assignment, preserving Canton's requirement for strictly increasing sequencing timestamps.

2.4 Governance Integration

The BFT ordering service fully integrates with Canton's existing topology management system. Synchronizer owners govern the ordering layer through standard Canton topology transactions, including adding/removing sequencers and configuring dynamic BFT parameters. Reconfiguration occurs at epoch boundaries; a notification service informs synchronizer owners when BFT ordering topology changes take effect.

The simplified sequencing model has all sequencer nodes participating in write, read, and ordering functions. Synchronizer owners are responsible for assessing fault tolerance implications of topology changes and setting the SequencerDomainStateX threshold accordingly.

2.5 Pruning and Persistence

The ordering service persists proofs of availability, active dynamic parameters, the most recent end-of-epoch transaction, partial consensus progress, the totally ordered log, and peer-to-peer connectivity state. Pruning proceeds iteratively across three phases: manual pruning by node operators, automatic distributed trustful pruning that coordinates across all ordering nodes, and finally automatic decentralized pruning tolerant of up to f Byzantine peers. Currently the Global Synchronizer prunes ordering service data after 30 to 31 days.

Critical safety rules govern pruning: batches must not be pruned before their expiration time (to prevent replay attacks via resubmitted proofs of availability), pruning operates at epoch granularity, and batches from removed ordering nodes are retained until natural expiration to preserve availability guarantees.

Milestones and Deliverables

Based on the execution plan, the project is divided into five milestones over an 8-month period.

Milestone 1: Core Primitives – Mempool, Availability & Persistence

Deliverable	Months after approval
Estimated Delivery	Month 2 (May 2026 if approved in March)
Focus	Building the foundational mempool and availability modules with persistence

Deliverables:

• Mempool module: queuing of ordering requests, batch creation on demand for the availability module.
• Availability module: reliable dissemination of request batches using proofs of availability (PoAs) with configurable quorum thresholds.
• Per-ordering-node storage fairness enforcement with configurable limits on unordered, unexpired PoAs.
• Batch expiration logic based on deterministic derivation from contained requests' max_sequencing_time.
• Garbage collection of expired batches at epoch boundaries.
• Persistence layer for PoAs, active dynamic parameters, and peer-to-peer connectivity state.
• Database schema design optimized for cross-module storage efficiency and pruning performance.
• Unit and integration tests for all mempool and availability module functionality.

Milestone 2: Parallel Ordering & BFT Timestamping

Estimated Delivery	Month 3 (June 2026 if approved in March)
Focus	ISS-based parallel consensus with deterministic timestamp assignment

Deliverables:

• ISS ordering module: epoch-based log subdivision with configurable epoch length (k slots) and segment count (s segments).
• Parallel consensus across segments, with each validator leading sequential instances within its assigned segment.
• Output module: deterministic transformation of ordered blocks, replacing batch references with full request data.
• Median-based BFT timestamp assignment using 2f+1 canonical signed protocol messages per validator block.
• Post-order timestamp shifting to maintain globally strictly increasing timestamps across interleaved parallel output.
• Per-transaction timestamp assignment at microsecond granularity with configurable block time granularity.
• Handling of edge cases: genesis/bootstrap blocks, end-of-epoch timestamp derivation, cancelled blocks (⊥ values).
• Integration test demonstrating parallel ordering across 4+ validators with consistent timestamp assignment and zero data loss.

Acceptance Criteria:
Verification of parallel consensus processing

Milestone 3: Governance, Security & Production Readiness

Estimated Delivery	Month 4 (July 2026 if approved in March)
Focus	Canton topology integration, governance workflows, security hardening, crash recovery

Deliverables:

• Full integration with Canton's topology management for BFT ordering governance: adding/removing sequencers, dynamic BFT parameter configuration via dedicated topology transaction types.
• Online reconfiguration at epoch boundaries: topology freeze, deferred activation, and BFT ordering topology notification service for domain owners.
• BFT domain bootstrap procedure: genesis package creation, P2P network formation with challenge-response authentication, and 2f+1 startup threshold.
• BFT sequencer onboarding workflow: state transfer, P2P integration at epoch change, and queuing/retry of incoming requests during synchronization.
• Manual pruning via dedicated Admin API gRPC endpoint, with operator documentation on safe pruning timestamps.
• Crash recovery: persistence of partial consensus progress, in-memory cache reconstruction for BFT time calculations, and consistency verification on restart.
• Security hardening: authenticated dissemination, protection against request flooding, replay attack prevention across epoch boundaries, and write relay support for observing nodes.
• Internal security audit covering BFT time monotonicity, deduplication correctness, and fault tolerance under topology changes.
• Final release: merged code into the main Canton branch, included in a tagged release.

Acceptance Criteria:

• Throughput Demonstration
• Governance Demonstration

Milestone 4: Automatic Pruning & Operational Maturity

Estimated Delivery	Month 5 (August 2026 if approved in March)
Focus	Automatic distributed pruning, operational tooling, documentation

Deliverables:

• Automatic distributed trustful pruning: ordering nodes periodically exchange highest consecutively completed block and epoch information; eligible blocks are pruned automatically once all nodes acknowledge delivery.
• Pruning safety enforcement: unexpired batches are retained regardless of acknowledgment status; pruning operates at epoch granularity; partial metadata retention for deduplication of unexpired proofs.
• Updated operator manuals covering BFT domain bootstrap, sequencer onboarding, governance examples, trust model and assumptions, and pruning configuration.
• Metrics and logging for all BFT ordering modules: mempool depth, availability dissemination latency, consensus progress per segment, timestamp drift, and pruning status.
• Code complete: All code merged into the open-source Canton repository with appropriate unit and integration test coverage, including fault injection tests for Byzantine scenarios.

Acceptance Criteria:

• Demonstrable fault tolerance: Demonstrate that performance doesn't drop when an ordering node drops out of the quorum.

Milestone 5: Successful Deployment on Global Synchronizer MainNet

Estimated Delivery	Month 8 (November 2026 if approved in March)
Focus	MainNet deployment achieving ≥50 tps sustained throughput

Deliverables:

• Documentation**:** Updated operator manuals covering BFT domain bootstrap, sequencer onboarding, governance procedures, trust model and assumptions, pruning configuration, and dynamic parameter tuning.
• Security Audit: Complete and deliver an internal security audit covering BFT time monotonicity, availability guarantees, deduplication correctness, storage fairness, and fault tolerance under topology changes.
• Deploy BFT ordering service on Global Synchronizer MainNet with at least 90 percent of Validators participating.
• Demonstrate sustained average throughput of at least 50 transactions per second under representative production load.
• Successfully execute at least one online BFT ordering reconfiguration (adding or removing a sequencer) on MainNet without perceptible downtime.
• Resolve any issues that arise during MainNet deployment and initial operation.

Acceptance Criteria:
MainNet Performance: Sustained average throughput of at least 50 Canton Coin-equivalent transactions per second (approximately 1 MB/sec burn) on the Global Synchronizer MainNet.

Operational Handover & Ongoing Maintenance

Upon the successful completion of Milestone 5, day-to-day maintenance of the ISS-based BFT repository (including security SLAs, bug fixes, CI/CD pipeline management, and external PR reviews) will seamlessly roll into the purview of Digital Asset’s proposed 2026-Maintenance Grant for Daml Open Source.

Note on Performance Targets

Digital Asset has built comprehensive simulations of the proposed BFT implementation, and has built extensive proofs of concept to prove out its viability. Based on the results of this work, we are confident that this protocol will scale over time to process on the order of 30,000 to 70,000 events per second.

A Canton Coin transfer requires events to process on approximately twenty Validators for every transaction. This maps to asset transfer rates on the order of 900 to 2,000 Canton Coin transfers per second, suggesting substantial long-term potential for this protocol.

Below is a table with rough scale calculations based on various factors: [source link]
![][image1]

Acceptance Criteria

The Tech & Ops Committee can evaluate the completion of the grant based on the following:

Code Quality: All code merged into the open-source Canton repository with appropriate unit and integration test coverage, including fault injection tests for Byzantine scenarios.

Documentation: Updated operator manuals covering BFT domain bootstrap, sequencer onboarding, governance procedures, trust model and assumptions, pruning configuration, and dynamic parameter tuning.

Security Audit: Complete and deliver an internal security audit covering BFT time monotonicity, availability guarantees, deduplication correctness, storage fairness, and fault tolerance under topology changes.

MainNet Performance: Sustained average throughput of at least 50 tps on the Global Synchronizer MainNet.

Funding

Total Funding Request: 20,000,000 Canton Coin (CC).

Payment Schedule:

Milestone	Amount (CC)	Trigger
1 – Core Primitives	1,500,000	Committee acceptance of deliverables
2 – Parallel Ordering	2,500,000	Committee acceptance of deliverables
3 – Governance & Release	2,500,000	Final release and committee acceptance
4 – Automatic Pruning	1,500,000	Committee acceptance of deliverables
5 – MainNet Deployment	8,800,000	Sustained ≥50 tps on MainNet

Early Completion Bonus:

We propose to complete feature work and testing within a month after MainNet adopts Logical Synchronizer Upgrades, and we plan to roll this feature out to MainNet within a month after that feature completion date. We further propose an Early Completion Bonus if this work is completed in time to be included as part of the first Logical Synchronizer upgrade on MainNet. .

Early Completion	Bonus Percentage	Bonus Amount
Completion together with the first Logical Synchronizer Upgrade on MainNet	20%	3,200,000

Volatility Stipulation for completion in more than 6 months

This grant is denominated in fixed Canton Coin. Should the project timeline extend beyond 6 months, the remaining un-minted milestones will be renegotiated to account for any significant USD/CC price volatility.

Co-Marketing

Upon the release of the BFT ordering service, Digital Asset will collaborate with the Foundation on:

• A technical blog post explaining the parallel ordering consensus architecture and its throughput benefits.
• Highlighting the feature in the quarterly Canton Development Fund report.

Rationale

The proposed architecture addresses each of the motivating concerns through four reinforcing design choices, grounded in the latest distributed systems research and proven in production-grade BFT systems.

Separating Dissemination from Consensus

The Narwhal paper demonstrated that removing data dissemination from the consensus critical path yields dramatically higher throughput, even when using non-DAG consensus algorithms such as HotStuff. By reliably disseminating transaction batches through the availability module before consensus begins, the ordering protocol operates on lightweight batch references rather than full transaction payloads. This reduces the communication overhead of each consensus round by orders of magnitude and makes throughput limited primarily by available network bandwidth for dissemination—which can be scaled independently—rather than by the consensus protocol itself.

Parallel Consensus via ISS

ISS divides the ordered log into epochs, with each epoch's slots distributed across segments led by different validators. This allows multiple validators to run independent consensus instances concurrently, producing blocks in parallel. Throughput scales with the number of validators rather than being constrained by a single leader. ISS also provides inherent ordering fairness: each validator contributes proportionally to the epoch's output, preventing any single validator from monopolizing ordering bandwidth.

Deterministic BFT Timestamping

The median-based timestamp approach, inspired by CometBFT but adapted for parallel consensus, ensures that block timestamps reflect the wall-clock times of correct validators even under Byzantine conditions. The median of 2f+1 signed messages guarantees that the selected timestamp was proposed by a correct node. Post-order shifting maintains strict monotonicity across the globally interleaved output, preserving Canton's invariant of strictly increasing sequencing timestamps without requiring tight clock synchronization. This approach enables all blocks from a given validator to proceed sequentially while allowing blocks from different validators to proceed in parallel, aligning with ISS's concurrency model.

Unified Governance via Canton Topology

By integrating BFT ordering governance with Canton's existing topology management system, this design avoids creating a separate governance plane for the ordering layer. Domain owners use the same authorization mechanisms, collective governance processes, and topology transactions they already understand. Reconfiguration at epoch boundaries, combined with the BFT ordering topology notification service, ensures that domain owners have the information they need to make safe governance decisions without requiring changes to Canton's topology system or the current trust model.

Iterative Pruning Toward Decentralization

The three-phase pruning strategy—manual, automatic trustful, and automatic decentralized—reflects a pragmatic approach to a hard distributed systems problem. Each phase delivers operational value while building toward the ideal end state. The critical safety invariants (batch expiration enforcement, epoch-granularity pruning, retention of unexpired batches) are established from the first phase and remain constant, ensuring that each phase transition does not introduce new correctness risks.

Design Alternatives Considered

The following alternative approaches were evaluated during the design process:

Approach	Advantages	Disadvantages	Decision
Full Narwhal DAG	Inherent storage fairness, proven in DAG BFTs, good latency (Shoal)	Higher implementation effort, wasted bandwidth on reconfiguration, fairness overkill for non-DAG consensus, design risk with ISS	Not selected
Home-grown PoA without DAG (selected)	Simpler implementation, better expected latency, ISS provides ordering fairness	Novel design risk, must implement independent fairness enforcement	Selected
CometBFT-like mempool gossip + vanilla ISS-PBFT	Best latency up to max throughput, no novel design risk	Significant mempool implementation, lower max throughput (~80k tx/s), can't leverage Narwhal fairness	Not selected
Vanilla ISS-PBFT with bucket sharding	Best latency, no novel design risk	Requires smarter clients, complex client-side logic, sequence number management, lower max throughput	Not selected

[image1]: <data:image/png;base64,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

Footnotes

1. When a sequencer receives a transaction, that sequencer is the only node that submits the transaction for ordering; it does not inform other nodes of transactions that are awaiting ordering: there is no pre-sharing of tx submissions across multiple nodes. ↩