IMMIX – Leveraging Aeron Cluster in digital assets trading

IMMIX – Leveraging Aeron Cluster for a sequencer-based trading system in digital assets trading

IMMIX is a pioneer in developing high-performance trading infrastructure for digital assets. At our Aeron MeetUp held in London in November 2024, Wenzhe Hu, Senior Software Engineer, and Andrew Mann, Co-Founder at IMMIX, shared their sequencer journey with Aeron. This blog post summarizes their talk, highlighting key points and inviting you to watch the full presentation for a deeper dive into architecture and benchmark data.

Who is IMMIX?

IMMIX focuses on scaling institutional digital asset trading through next-generation technologies. Their platform offers a suite of optimal execution algorithms designed for precision and efficiency, as well as market-making strategies aimed at enhancing digital asset liquidity.

Challenges in crypto / digital asset trading

The crypto trading landscape presents unique challenges:

• 24/7 operation: Unlike traditional markets, crypto markets operate non-stop, requiring zero downtime models.
• Fragmented global market: Numerous exchanges and liquidity providers are spread across different locations leading to market inefficiencies.
• Fast-moving space: Rapid development cycles are essential to keep up with the fast-paced market.
• Primarily cloud-based: Operating on the cloud introduces less stable environments and unpredictable latency.

IMMIX’s sequencer and its design

To address these challenges, IMMIX has implemented a sequencer-based system using Aeron Cluster. Here’s a breakdown of the sequencer design and workflow:

Sequencer workflow:

1. Active Client App: Publishes messages via cluster ingress.
2. Passive Client App: Tracks in-flight messages without sending them to the sequencer.
3. Sequencer Nodes: Validate and sequence messages, publishing them over a recorded IPC channel.
4. Local IPC Replay: All apps replay messages via local IPC.
5. Remote Client Support: Requests replay from the sequencer.
6. Leader Node Replay: Starts replay and responds with the replay session ID.

The system is designed to handle leader and client failovers seamlessly, ensuring data consistency and minimizing data loss.

Advantages of using Aeron Cluster

The Aeron Cluster-based sequencer offers several advantages over traditional active/passive setups:

• Fault tolerance: Configurable resilience with multiple nodes.
• Performance: Maintains low latency crucial for trading.
• Throughput: Supports multi-inflight messages, ensuring high throughput.
• Consistency: Strong consistency with consensus-based replication.
• Development: Easier local setup and debugging, including failover scenarios.

Additionally, the benchmarks discussed in the presentation highlight the system’s impressive performance, demonstrating its capability to handle high message rates with minimal latency.

Use Case: Quant research and real-time experimentation

At the end of the presentation, Andrew Mann discusses how quant research teams can benefit from a sequencer architecture. Discussion topics include…

• Feature commoditization: Allowing for rapid iteration and deployment of models in real-time.
• Real-time experimentation: Enabling monitoring and analysis of deeper layers within machine learning models without impacting trading performance.
• Model coordination: Facilitating communication between different models, enhancing the overall trading strategy.

IMMIX’s innovative use of Aeron Cluster for its sequencer system addresses the unique challenges of crypto trading, providing a robust, high-performance solution. The detailed diagrams and benchmark data presented in the talk offer valuable insights into the system’s design and performance.

About IMMIX

Wenzhe Hu
Senior Software Engineer
IMMIX

Wenzhe is a lead engineer at IMMIX who has worked in several front-office divisions at major investment banks such as UBS and crypto exchange Crypto.com. He has an MRes in Photonics from the University of Cambridge.

 

Andrew Mann
Co-Founder
IMMIX

Andrew is a co-founder at IMMIX and has previously worked at several institutions and quantitative hedge funds including Morgan Stanley, Matrix8J, DV Trading and Virtu Financial as a front office quantitative researcher. He holds several engineering degrees with the most recent being a PhD from the Computer Science department at UCL.

 

IMMIX is an institutional cloud trading platform for digital assets, providing advanced trading tools, risk management, and market access for hedge funds, brokers, and financial institutions. Supporting over 25 exchanges and 3,000+ instruments, it enables seamless 24/7 trading with high-performance infrastructure. Backed by MassMutual Ventures and Ripple, IMMIX has facilitated over $6 billion in traded volume, positioning itself as a key player in institutional digital asset trading.