|
1 | 1 | ---
|
2 |
| -title: "Consensus" |
| 2 | +title: Consensus in TON |
| 3 | +description: Deep technical overview of the consensus protocols in the TON Blockchain, including Catchain, Block Consensus Protocol, validator elections, incentives, and fault tolerance. |
3 | 4 | ---
|
4 | 5 |
|
5 |
| -Stub |
| 6 | +## Introduction |
| 7 | + |
| 8 | +The TON Blockchain achieves consensus through a layered Byzantine Fault Tolerant (BFT) protocol designed for **high throughput, low latency, and security against up to one-third malicious validators**. |
| 9 | + |
| 10 | +The system combines two protocols: |
| 11 | + |
| 12 | +- **Catchain Protocol** – a reliable broadcast mechanism that ensures consistent message delivery and fork detection. |
| 13 | +- **Block Consensus Protocol (BCP)** – a three-phase commit protocol built on top of Catchain that finalizes blocks. |
| 14 | + |
| 15 | +This separation allows TON to finalize blocks in **3–6 seconds** with **hundreds of validators** across the globe. |
| 16 | + |
| 17 | +--- |
| 18 | + |
| 19 | +## Consensus Model |
| 20 | + |
| 21 | +### Byzantine Fault Tolerance |
| 22 | +Let: |
| 23 | +- `n` = number of validators, |
| 24 | +- `f` = number of faulty or malicious validators, |
| 25 | +- `q` = quorum size. |
| 26 | + |
| 27 | +Conditions for TON consensus: |
| 28 | + |
| 29 | +``` |
| 30 | +f < n/3 |
| 31 | +q ≥ 2n/3 |
| 32 | +``` |
| 33 | + |
| 34 | +These conditions guarantee: |
| 35 | +- **Safety** – no two conflicting blocks can both be finalized. |
| 36 | +- **Liveness** – as long as ≥ `2n/3` validators are responsive, some block is always finalized. |
| 37 | + |
| 38 | +--- |
| 39 | + |
| 40 | +## Catchain Protocol |
| 41 | + |
| 42 | +Catchain provides the secure communication foundation for BCP. |
| 43 | + |
| 44 | +### Validator messages |
| 45 | +Each validator `v` generates a sequence of signed messages: |
| 46 | + |
| 47 | +``` |
| 48 | +m(v, h) = (id = v, height = h, deps[], payload, sig) |
| 49 | +``` |
| 50 | + |
| 51 | +- `height`: strictly increasing local counter. |
| 52 | +- `deps[]`: references to previous messages (DAG edges). |
| 53 | +- `sig`: cryptographic signature binding the message. |
| 54 | + |
| 55 | +### DAG structure |
| 56 | +The global state of Catchain is a **Directed Acyclic Graph (DAG)** where: |
| 57 | +- Vertices are messages. |
| 58 | +- Edges represent dependencies. |
| 59 | +- Every message must reference known predecessors. |
| 60 | + |
| 61 | +This enforces causal ordering: a message can only be processed when its entire dependency cone is available. |
| 62 | + |
| 63 | +### Fork detection |
| 64 | +Forking occurs when: |
| 65 | + |
| 66 | +``` |
| 67 | +∃ m1, m2 : m1.id = m2.id, m1.height = m2.height, m1 ≠ m2 |
| 68 | +``` |
| 69 | + |
| 70 | +This is undeniable evidence of equivocation. |
| 71 | +A **fork proof** = `{m1, m2}` can be submitted to the Elector contract, leading to validator slashing. |
| 72 | + |
| 73 | +### Guarantees |
| 74 | +- **Consistency** – all honest validators eventually converge on the same DAG. |
| 75 | +- **Integrity** – equivocation is always detectable. |
| 76 | +- **Foundation** – BCP builds on this consistent DAG to safely finalize blocks. |
| 77 | + |
| 78 | +--- |
| 79 | + |
| 80 | +## Block Consensus Protocol (BCP) |
| 81 | + |
| 82 | +With Catchain providing reliable messaging, BCP ensures validators agree on block finalization. |
| 83 | + |
| 84 | +### Phases |
| 85 | +BCP is structured as a **three-phase commit**: |
| 86 | + |
| 87 | +1. **Proposal** – a leader proposes a block candidate. |
| 88 | +2. **Validation** – validators verify correctness (transaction validity, state consistency). |
| 89 | +3. **Voting** – validators vote for approved candidates. |
| 90 | +4. **PreCommit** – once ≥ `q` votes are collected, validators broadcast PreCommit. |
| 91 | +5. **Commit** – once ≥ `q` PreCommits are observed, validators issue CommitSign and the block is finalized. |
| 92 | + |
| 93 | +If no block reaches quorum, a **null block** is finalized, guaranteeing progress. |
| 94 | + |
| 95 | +### Safety proof |
| 96 | +Suppose two conflicting blocks `B1` and `B2` are both finalized. |
| 97 | +- Each requires ≥ `q ≥ 2n/3` signatures. |
| 98 | +- Their signer sets intersect in ≥ `2q – n ≥ n/3` validators. |
| 99 | +- Since `f < n/3`, at least one honest validator signed both, which is impossible. |
| 100 | + |
| 101 | +Therefore, conflicting blocks cannot both be finalized. |
| 102 | + |
| 103 | +### Liveness |
| 104 | +As long as ≥ `2n/3` validators are online and responsive, some block or null block always gathers enough signatures, ensuring bounded progress. |
| 105 | + |
| 106 | +--- |
| 107 | + |
| 108 | +## Rounds and Attempts |
| 109 | + |
| 110 | +Consensus is organized into **rounds** and **attempts**: |
| 111 | + |
| 112 | +- Each round lasts one election cycle (~18h on mainnet, ~2h on testnet). |
| 113 | +- Each round is subdivided into attempts of fixed duration (`K ≈ 8s`). |
| 114 | +- **Fast attempts** – optimistic, finalize quickly if leader is honest and network healthy. |
| 115 | +- **Slow attempts** – coordinator-driven, ensure progress under failures. |
| 116 | + |
| 117 | +Block latency is typically **3–6s**, bounded by attempt timers. |
| 118 | + |
| 119 | +--- |
| 120 | + |
| 121 | +## Validator Elections |
| 122 | + |
| 123 | +Validators are selected by the **Elector contract** using an on-chain election. |
| 124 | + |
| 125 | +### Process |
| 126 | +1. Validators submit stake transactions. |
| 127 | +2. Elector sorts candidates by stake. |
| 128 | +3. Up to `maxValidators` are selected. |
| 129 | +4. Effective stake is capped to ensure fairness. |
| 130 | + |
| 131 | +### Effective stake formula |
| 132 | +``` |
| 133 | +effectiveStake(v) = min(stake(v), minStake × stakeFactor) |
| 134 | +``` |
| 135 | + |
| 136 | +Where: |
| 137 | +- `minStake` = minimum stake among selected validators, |
| 138 | +- `stakeFactor` ≈ 3. |
| 139 | + |
| 140 | +This prevents one validator from dominating with excessive stake. |
| 141 | + |
| 142 | +### Example |
| 143 | +If the smallest validator stakes 100k TON, then: |
| 144 | +- The largest counted stake = 300k TON. |
| 145 | +- A validator staking 1M TON still only contributes 300k TON effectively. |
| 146 | + |
| 147 | +### Set sizes |
| 148 | +- **Masterchain** – ~100 validators. |
| 149 | +- **Shardchains** – ~23 validators each. |
| 150 | + |
| 151 | +--- |
| 152 | + |
| 153 | +## Incentives and Penalties |
| 154 | + |
| 155 | +### Rewards |
| 156 | +Validators are compensated with: |
| 157 | +- **Transaction fees** (gas costs). |
| 158 | +- **Block subsidies**: |
| 159 | + - ~1.7 TON per masterchain block. |
| 160 | + - ~1 TON per shardchain block. |
| 161 | + |
| 162 | +Average income: ~120 TON per validator per round (varies with network load). |
| 163 | + |
| 164 | +### Inflation and burn |
| 165 | +- Inflation rate: ~0.3–0.6% annually. |
| 166 | +- Since June 2023, part of the subsidy is burned, introducing deflationary pressure as usage grows. |
| 167 | + |
| 168 | +### Penalties |
| 169 | +Validators may be fined for: |
| 170 | +- **Inactivity** – not producing or signing enough blocks (lower than 90% efficiency). |
| 171 | +- **Malicious behavior** – forks, equivocation, invalid approvals. |
| 172 | +- **Standard fine** – ~101 TON per round of misbehavior. |
| 173 | + |
| 174 | +### Slashing mechanism |
| 175 | +- Evidence is submitted as fork proofs or efficiency complaints. |
| 176 | +- Validators verify collectively; ≥ `2n/3` agreement required. |
| 177 | +- Elector contract enforces slashing automatically. |
| 178 | + |
| 179 | +This ensures accountability while preventing abuse by a small minority. |
| 180 | + |
| 181 | +--- |
| 182 | + |
| 183 | +## Validator Guidelines |
| 184 | + |
| 185 | +Operating a validator reliably is critical. Recommended practices: |
| 186 | +- Run on high-performance, redundant servers. |
| 187 | +- Ensure stable, low-latency network connections. |
| 188 | +- Use monitoring systems (Prometheus, Grafana, Datadog) to track CPU, memory, disk, and validator efficiency. |
| 189 | +- Keep validator software updated to the latest stable release. |
| 190 | +- React quickly to alerts about downtime or forks. |
| 191 | + |
| 192 | +Validators unable to meet these requirements may prefer to delegate stake via staking services. |
| 193 | + |
| 194 | +--- |
| 195 | + |
| 196 | +## Extended Fault Tolerance Analysis |
| 197 | + |
| 198 | +- **Fault tolerance:** |
| 199 | + - With `n` validators, up to `f < n/3` may behave arbitrarily (Byzantine). |
| 200 | + - Finalization requires ≥ `2n/3` CommitSigns. |
| 201 | + |
| 202 | +- **Intersection property:** |
| 203 | + - Any two quorums of size ≥ `2n/3` intersect in at least `n/3`. |
| 204 | + - Since `f < n/3`, intersection always includes an honest validator. |
| 205 | + - This guarantees that conflicting blocks cannot both finalize. |
| 206 | + |
| 207 | +- **Efficiency requirement:** |
| 208 | + - A validator’s efficiency = `signedBlocks / expectedBlocks`. |
| 209 | + - If efficiency is lower than 90% in a round, fines are applied. |
| 210 | + |
| 211 | +- **Liveness bound:** |
| 212 | + - Consensus always finalizes within a bounded number of attempts. |
| 213 | + - Null blocks prevent deadlock even under heavy network partitions. |
| 214 | + |
| 215 | +--- |
| 216 | + |
| 217 | +## Summary |
| 218 | + |
| 219 | +- **Catchain Protocol** – consistent broadcast of validator messages with DAG structure and fork proofs. |
| 220 | +- **Block Consensus Protocol** – three-phase commit finalizing blocks with ≥ `2n/3` quorum. |
| 221 | +- **Elections** – validator selection via the Elector contract, with fairness enforced by stake caps. |
| 222 | +- **Rewards and penalties** – align incentives with network security. |
| 223 | +- **Fault tolerance** – safety holds with `f < n/3`, liveness guaranteed with ≥ `2n/3` honest validators. |
| 224 | + |
| 225 | +This architecture enables TON to deliver **fast finality, decentralization, and strong BFT guarantees** at global scale. |
0 commit comments