Introduction

In the fast-paced world of high-frequency trading (HFT), every microsecond counts. Today, I'll share how to implement a dual Ultra-Low Latency (ULL) connection setup that's becoming crucial for trading operations, particularly when connecting to cloud infrastructure like AWS.

‍

Why This Matters

For traders, especially those dealing with algorithmic and high-frequency trading, traditional cloud connectivity often isn't fast enough. Cloud providers' native backbones can introduce unnecessary latency. By implementing a dual ULL setup, we can:

• Achieve sub-millisecond latency
• Ensure 100% uptime through redundancy
• Bypass slower cloud provider backbones
• Maintain consistent performance during network issues

The Architecture

Our setup involves:

• Two separate ULL subsea connections
• Four Direct Connect (DX) locations (two per region)
• BGP Active/Standby configuration
• Multiple Virtual Interfaces (VIFs)

Key Components:

• 2 x DX Gateways
• 4 x DX Connections (2 per region)
• 4 x VIFs
• Each DXGW associated with 2 x VIFs & 1 x VGW

The BGP Setup

We implement an Active/Standby BGP configuration where:

• Primary path uses DX Location 1
• Secondary path uses DX Location 2 with BGP prepend
• Both paths utilize separate ULL subsea cables
• BFD (Bidirectional Forwarding Detection) enabled for rapid failover

‍

Failover Mechanism

The system provides seamless failover through:

• Active BGP path monitoring
• Automatic failover to secondary path
• Zero packet loss during transition
• Automatic recovery when primary path returns

Failover Test 1 - Simulation on Primary ULL Path BGP down

On AWS Console, select the "SG DX Location 1 vif" to simulate the "ULL Subsea 1" connection down.

We continuously performed ping tests between the Singapore EC2 and Tokyo EC2 instances, and the failover results demonstrated 0% packet loss while switching to the backup path.

‍

The results also demonstrated that this dual BGP setup can maintain AWS routing, ensuring the workload remains unaffected and the failover is extremely fast.

‍

Failover Test 2 - Simulation on Primary ULL "Subsea" cable breaks

This time we break the connection between Singapore and Tokyo routers

‍

‍

The results below illustrate the routing path before and after the failure of ULL Path 1. They confirm that no packet loss occurred during the transition, as traffic seamlessly rerouted to ULL Path 2 without any interruptions. Most importantly, the dual-link setup with dual BGPs and DXGW demonstrated flawless failover functionality, as expected.

‍

‍

Conclusion

While this setup requires significant investment and expertise, it's becoming a standard requirement for serious trading operations. The combination of ultra-low latency and high resilience provides the infrastructure backbone needed for modern trading operations.

Remember: in trading, being just a millisecond faster can mean the difference between profit and loss. This dual ULL setup ensures you're always operating at peak performance, with a backup ready to take over instantly if needed.

If your team needs our Ultra Low Latency network expertise or would like to discuss deployment options, our specialists are ready to help. Reach out to us at info@a-fieldtech.com to explore how we can enhance your trading infrastructure with our proven ULL solutions.

‍