Design and Implementation of a Scalable Financial Exchange in the Public Cloud

2024-02-14

AI Summary

Public cloud “best-effort” networks fail to meet financial exchanges’ sub-microsecond latency and high-throughput order-processing requirements. To address this, we propose Jasper—a low-latency, high-throughput trading infrastructure for public clouds. Our approach introduces three key innovations: (1) an overlay tree architecture that unifies market-data multicast and order reverse ingress; (2) a limit-order queue (LOQ) coupled with dynamic order throttling to enhance throughput under bursty workloads; and (3) synergistic optimizations including kernel bypass, priority-based scheduling, and precision clock synchronization. Evaluation shows Jasper achieves end-to-end latency of 250 μs and inter-node data-reception skew ≤1 μs. Compared to AWS Elastic Fabric Adapter (EFA) multicast, Jasper reduces latency by 50% and improves burst-order throughput by 97%.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Distributed Search
• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web
• Security and Privacy: Large-scale security measurements

Abstract

Financial exchanges are migrating to the cloud, but the best-effort nature of the public cloud is at odds with the stringent latency requirements of exchanges. We present Jasper, a system for meeting the networking requirements of financial exchanges on the public cloud. Jasper uses an overlay tree to scalably multicast market data from an exchange to ~1000 participants with low latency (250 microseconds) and a 1-microsecond difference in data reception time between any two participants. Jasper reuses the same tree for scalable inbound communication (participants to exchange), augmenting it with order pacing and a new priority queue, Limit Order Queue (LOQ), to efficiently handle bursts of market orders. Jasper achieves better scalability and 50% lower latency than the AWS multicast service. During bursty market activity, LOQ nearly doubles the order processing rate.

Problem

• Cloud Computing
• Financial Market
• High Performance Computing

Innovation

• Tree-based Networking
• Order Throttling
• Limit Order Queue Optimization