Narwhal DAG: High-Throughput BFT Consensus

• Narwhal DAG is a directed acyclic graph-based architecture that underpins Byzantine Fault Tolerant consensus, enabling rapid data broadcast and reliable transaction ordering.
• It decouples data dissemination from consensus ordering, ensuring strong causal consistency, fault tolerance through quorum-based protocols, and high throughput scalability.
• Implementations like Bullshark, Shoal, Rorqual, MRV, and Herring demonstrate ongoing research in fair ordering, latency reduction, and TEE acceleration within Narwhal-based consensus stacks.

A Narwhal DAG (“Directed Acyclic Graph”) is a state-of-the-art Byzantine Fault Tolerant (BFT) broadcast and mempool abstraction foundational to many high-throughput, low-latency blockchain consensus protocols. Its design decouples disseminating transactional data from later agreement and ordering, enabling rapid wire-speed ingest while providing strong availability and causality guarantees. This architecture, originating from the Narwhal protocol (Danezis et al., 2021), underpins a family of BFT consensus stacks including Bullshark (Tanaka, 7 Jul 2025), Shoal (Spiegelman et al., 2023), and Tusk, and is subject to active research on both the data plane (broadcast and DAG-building) and the ordering/consensus plane (e.g., post-consensus fairness, parallel ordering, TEE acceleration).

1. Round-Based DAG Formation in Narwhal

The Narwhal DAG is constructed in rounds, with each validator (primary) operating a local view of the DAG $D_p(t)$ at time $t$ (Danezis et al., 2021, Tanaka, 7 Jul 2025). Each vertex $v$ in round $r(v)$ encodes: (1) a batch of client transactions $\mathsf{txs}(v)$, (2) its creator $o(v)$, and (3) a set of parent-edges to at least $n-f$ round-$(r-1)$ vertices, ensuring every new vertex references a quorum of the immediately prior round (Spiegelman et al., 2023).

The structure is defined recursively: every new certified block or vertex—carrying signatures from a quorum ($2f+1$ or $n-f$) of validators—references its parent set, thereby encoding a “happens-before” relation. This parent linkage ensures that the sub-DAG $t$0 reachable from $t$1 captures complete causal history, and the non-equivocation property is enforced by limiting honest primaries to one vertex per round (Spiegelman et al., 2023).

DAG Growth Rule

Validators advance rounds by collecting at least $t$2 round-$t$3 vertices, then broadcasting exactly one vertex for round $t$4, referencing any $t$5 round-$t$6 vertices delivered so far. This rule ensures:

• Validity: If an honest validator has $t$7 then all of $t$8 is present.
• Eventual Delivery: If an honest validator has $t$9, all honest validators eventually receive $v$0.
• Non-equivocation: Honest validators broadcast at most one vertex per round.

Mathematically, the vertex and edge set per round $v$1 are:

$v$2

Quorum intersection properties guarantee, for $v$3, that any two sets of size $v$4 intersect in at least $v$5 honest nodes (Danezis et al., 2021).

2. DAG-Based Reliable Broadcast and Mempool Invariants

Narwhal separates data dissemination from ordering by designing a robust round-based DAG whose only aim is to provide:

• Data availability: Each batch is stored at $v$6 nodes, guaranteeing future fetch-ability.
• Causal consistency and integrity: Containment and ancestry queries are deterministic due to the closure of parent edge references.
• Scale-out capability: Workers collect, batch, and reliably broadcast data in parallel, while primaries handle only compact metadata (Danezis et al., 2021, Tanaka, 7 Jul 2025).

Garbage collection is integrated via an agreed GC-round, deleting DAG vertices whose rounds are less than $v$7 and thus bounding memory consumption without risking causality for future ordering (Tanaka, 7 Jul 2025).

Empirically, with 4 to 10 workers per validator, throughput scales quasi-linearly to hundreds of thousands of tx/s, e.g., 600,000 tx/s at sub-3 s latency for 5 workers ×50 validators (Danezis et al., 2021).

3. Consensus Mechanisms Supported by the Narwhal DAG

Bullshark: Zero-Overhead, Round-Based DAG Consensus

Bullshark orders transactions by interpreting the Narwhal DAG as an implicit voting structure (Spiegelman et al., 2023, Tanaka, 7 Jul 2025). The protocol alternates anchor rounds (even, leader-broadcast) and vote rounds (odd, all validate anchors). Commitment uses an $v$8-vote rule: once a validator sees $v$9 round-$r(v)$0 vertices voting for an anchor $r(v)$1, $r(v)$2 is committed; all vertices in its causal sub-DAG $r(v)$3 are then output via a deterministic order, ensuring agreement by DAG completeness.

Formally:

$r(v)$4

Bullshark achieves throughput up to 297,000 tx/s at 2 s latency on a 50-node deployment (Tanaka, 7 Jul 2025).

Shoal: Pipelining, Leader-Reputation, Prevalent Responsiveness

Shoal generalizes any Narwhal-based consensus by pipelining Bullshark “instances” and introducing a dynamic leader-reputation scheme. Pipelining means every round is effectively an anchor round, reducing worst-case ordering latency for non-anchors from 4 to 3 rounds. The leader-reputation mechanism demotes persistently slow/crashed nodes, auto-updating the leader selection function $r(v)$5 based on the observable causal history.

Shoal exhibits prevalent responsiveness: honest replicas can commit without explicit timeouts except in pathologically rare cases—improving upon optimistic responsiveness guarantees of prior BFT protocols. Experimentally, Shoal reduces latency by up to 40% in the failure-free case and up to 80% under failure, with zero throughput penalty (Spiegelman et al., 2023).

4. Broadcast Acceleration and Security: Rorqual and TEEs

Rorqual enhances the Narwhal DAG-mempool using Trusted Execution Environments (TEEs) to collapse Narwhal’s multi-step reliable broadcast into a single message delay. The enclave signs, erasure-codes, and disperses vertices in one step, halving certificate and payload latencies ($r(v)$6 good and bad), mitigating equivocation, and providing accountability for omission faults (delays are logged and exposed by the TEE) (Freitas et al., 2024).

Key comparative metrics:

Protocol	Certificate Latency	Payload Latency
Narwhal Pull	$r(v)$7/3$r(v)$8	$r(v)$9/5$\mathsf{txs}(v)$0
Rorqual	1$\mathsf{txs}(v)$1/1$\mathsf{txs}(v)$2	1$\mathsf{txs}(v)$3/1$\mathsf{txs}(v)$4

When integrated into consensus stacks (e.g., Bullshark), Rorqual can yield $\mathsf{txs}(v)$5 end-to-end latency reduction and enables mixed-mode deployments (TEE/non-TEE) simply by falling back to pull-based recovery for non-TEE participants (Freitas et al., 2024).

5. Fair Ordering and Post-Consensus Ordering Layers

Multi-Round Visibility (MRV)

MRV addresses fair ordering by analyzing post-commit DAG slices to derive structural precedence evidence. For concurrently committed units, it tracks their visibility across up to $\mathsf{txs}(v)$6 rounds, inferring if one unit “reaches” a $\mathsf{txs}(v)$7 majority of creators ahead of another (structural visibility precedence, SVP). MRV produces a total order per slice, enforcing all causality and SVP-backed constraints, with deterministic tie-breaking for ambiguities (Ren et al., 22 May 2026). MRV adds negligible throughput cost (identical peak TPS) and low bounded latency overhead ($\mathsf{txs}(v)$8 ms in typical cases).

Herring: Parallel Batch-Order-Fairness

Herring parallelizes fair ordering graph construction across committed subdags, combining explicit missing-edge resolution piggybacked on DAG reliable-broadcast. It achieves $\mathsf{txs}(v)$9-batch-order-fairness, ensuring that if at least $o(v)$0 replicas receive $o(v)$1 before $o(v)$2, then $o(v)$3 must be ordered no later than $o(v)$4 ($o(v)$5 recovers strongest fairness). By parallelizing the dominant per-batch computation and resolving missing edges “off-path,” Herring provides up to $o(v)$6 higher saturation throughput than previous fair-ordering DAG protocols at $o(v)$7 (Putnik et al., 22 May 2026).

6. Trade-Offs, Performance, and Fault Tolerance

Narwhal’s separation of dissemination and agreement maximizes throughput and parallelism but introduces nuanced CAP trade-offs:

• Consistency & Partition Tolerance are ensured via deterministic total order and quorum-based causality.
• Availability is best-effort in periods of asynchrony or extended partition; fairness can be sacrificed by aggressive garbage collection (old blocks dropped).
• Latency is improved via Shoal (pipelining, leader-reputation) and Rorqual (TEE broadcast).
• Fault Tolerance: Quorum intersection and broadcast rules guarantee safety (non-equivocation, availability) up to $o(v)$8 Byzantine faults. Failures primarily affect latency, not throughput.

Empirically, Shoal+Bullshark achieves $o(v)$9 k TPS at $n-f$0–$n-f$1 ms latency; Rorqual further reduces ordering latency by up to 20%; MRV and Herring keep post-consensus order/cpu cost well below network bottlenecks (Spiegelman et al., 2023, Freitas et al., 2024, Ren et al., 22 May 2026, Putnik et al., 22 May 2026).

7. Extensions, Ongoing Research, and Comparative Landscape

Narwhal DAGs continue to serve as the backbone for advanced BFT stack research, including:

• Fully asynchronous consensus (Tusk/DAG-Rider)
• TEE-accelerated broadcast (Rorqual)
• Fair ordering with post-consensus modules (MRV, Herring)
• Application-specific optimizations (e.g., object-centric models in Sui)

Comparative studies consistently show Narwhal-based stacks outperforming leader-based (e.g., HotStuff, Themis) and in-graph fairness designs (e.g., FairDAG-RL, DoD) both in throughput and failure robustness.

Future directions include dynamic fairness/capacity trade-offs, further acceleration via hardware, and deeper exploration of composable fairness/evidence layers over authenticated DAGs.

---

Key References:

• "Shoal: Improving DAG-BFT Latency And Robustness" (Spiegelman et al., 2023)
• "Rorqual: Speeding up Narwhal with TEEs" (Freitas et al., 2024)
• "Narwhal and Tusk: A DAG-based Mempool and Efficient BFT Consensus" (Danezis et al., 2021)
• "Bullshark on Narwhal: Implementation-level Workflow Analysis..." (Tanaka, 7 Jul 2025)
• "Multi-Round Visibility: A Post-Consensus Ordering Layer..." (Ren et al., 22 May 2026)
• "Herring: Parallel Batch-Order-Fairness on DAG-based Blockchain Consensus" (Putnik et al., 22 May 2026)