Consider: when a major exchange suddenly goes down — REST API returns 503, WebSocket disconnects, orders get stuck. A 24/7 trading bot goes blind: no new trades, positions unmonitored. In one hour of downtime, a high-frequency strategy can lose up to $50,000 in potential profit. Without a failover system, such outages cost tens of thousands of dollars each time.
We design and implement turnkey failover systems — an automatic mechanism to switch operations to a backup exchange upon failure (failover). With over 5 years in DeFi and HFT and more than 50 completed projects, our solutions guarantee 99.9% uptime even if the primary platform fails. For clients with turnover exceeding $10M, average savings from implementation reach $200,000 per year. Failover investments pay back within months by preventing downtime.
Failover Architectures: Active-Passive and Active-Active
Two approaches: Active-Passive and Active-Active. In the first, the primary exchange handles all orders; the backup stays on standby. On primary failure, switching occurs. Simple, no duplicate orders. In the second, trading runs on multiple exchanges simultaneously; when one fails, others continue. More complex due to position coordination. For most bots, Active-Passive suffices.
| Criterion | Active-Passive | Active-Active |
|---|---|---|
| Implementation complexity | Low | High |
| Capital usage | Medium (two exchanges) | High (multiple exchanges) |
| Risk of duplicate orders | Minimal | Requires coordination |
| Uptime on single exchange failure | 99.9% | 99.99% |
Active-Passive is 2x simpler than Active-Active and requires 50% less capital. If your bot uses arbitrage or statistical advantage on one exchange, Active-Passive is optimal. For high-frequency trading with liquidity distribution, Active-Active is better, though it's more complex and costly.
Why Failover Is Critical for DeFi and HFT?
DeFi protocols run on smart contracts that depend on oracle data and network availability. If an exchange stops processing orders, arbitrage opportunities vanish and positions may get liquidated. In HFT, every millisecond of downtime means lost trades. A failover system ensures continuity even during planned maintenance or sudden outages.
How to Detect Exchange Failure?
Failure is not binary. Gradations: REST API unavailable, WebSocket disconnected, API responds but orders don't go through, latency increased 10x. Health check should monitor a combination of indicators: ping, market data, test order. Trigger threshold based on aggregate, not a single signal. Flapping protection: after switchover, a cooldown of 5–15 minutes prevents bouncing between exchanges during instability.
What to Do with Open Positions During Failover?
Three options: leave positions on the primary (risk without monitoring), mirror hedging (open opposite positions on backup to create net neutral exposure), or pause (no new positions, wait for recovery). Choice depends on strategy type and risk tolerance. We help find the balance between continuity and risk.
How to Set Up a Failover System: Step by Step
- Analyze strategy: identify activities vulnerable to downtime.
- Choose architecture: Active-Passive for simplicity, Active-Active for HFT.
- Configure health checks: integrate REST and WebSocket checks with thresholds.
- Implement flapping protection: set cooldown of 10 minutes.
- Integrate backup exchange: configure API keys, balances, fee structures.
- Test: simulate primary exchange failure and verify switchover.
- Deploy and monitor: roll out with logging and alerts.
Case Study
For a client running market making on Binance, we implemented Active-Passive failover with OKX as backup. Under normal operation, 100% of orders went to Binance. Over two months of operation, only one failover event occurred: Binance was down for 30 seconds due to an unscheduled update. During that time, the backup exchange processed 1,200 orders without any losses. Thanks to flapping protection, the system did not switch back immediately after recovery but observed the cooldown, eliminating unnecessary oscillations. As a result, uptime reached 99.95%, and the client avoided losses that could have amounted to tens of thousands of dollars.
Practical Limitations
Capital must be maintained on both exchanges — locking up funds. Prices of the same pair may differ — a strategy with tight levels may yield different results. Fee structures vary: 0.1% on one exchange, 0.15% on another — profit changes. Our engineers account for these nuances when designing, selecting the optimal pair of exchanges. Requirements for the backup exchange: support for the same set of trading pairs (or with minimal differences), API with comparable limits and stability, fee structure close to the primary to avoid strategy distortion.
What Is Included (Deliverables)
- Architectural documentation with rationale for exchange selection and failover scheme.
- Implementation of integration code on Foundry with unit tests.
- Configuration of health checks and flapping protection.
- Operation manual for your team.
- Staff training (1 hour online).
Work Process
| Stage | Duration | Result |
|---|---|---|
| Strategy analysis | 1–2 days | Requirements specification |
| Failover design | 2–3 days | Architecture documentation |
| Implementation on Foundry | 3–5 days | Code, tests, CI/CD |
| Exchange integration | 1–2 days | API connection |
| Testing | 2–3 days | Load test report |
| Deployment and documentation | 1 day | Manual, training |
We guarantee reliability: the system undergoes formal testing and operates under load. Experience: 50+ projects, including for market makers with multi-million dollar turnovers.
Order development of a failover system tailored to your strategy — we will analyze your requirements and propose the optimal solution. Contact us for a consultation today.