Inside the Securities Tech Revolution: Cloud, Microservices, and Big Data

Industry Paradox

The securities sector in China suffers from legacy IT that offers poor user experience, yet the same industry demands high‑performance quantitative and high‑frequency trading systems, creating a paradox between business innovation needs and technological stagnation.

Team Background and Philosophy

Since 2013 the team has operated as a FinTech startup, adopting aggressive technologies such as AngularJS, Docker, and Node.js. They position themselves as an R&D‑focused group that values open‑source tools, MacBook workstations, and a culture of rapid experimentation.

Technology Stack

The core stack consists of Node.js for high‑concurrency, single‑threaded, message‑driven services, and MongoDB for flexible, schema‑less storage of diverse financial products. The team avoids costly Oracle solutions, preferring open‑source alternatives to handle large‑scale data.

Trading Cloud Platform

The company built the only broker‑level trading cloud platform in China, offering API‑driven market data, simulation, and direct trading capabilities. The platform aims to create a developer ecosystem similar to Apple’s or Google’s, encouraging third‑party integration via well‑documented APIs.

Microservice Architecture

To modernize legacy systems, a “thick” service layer was introduced, employing service registration and discovery, contract‑driven testing (PACT), and automated API documentation generation. Tools such as Consul or etcd support service discovery, while the architecture emphasizes robustness, testability, and graceful degradation to avoid “silent death” of services during trading.

Big Data Pipeline

Log collection is performed with Flume and Kafka, feeding into Hadoop and Spark clusters for batch and near‑real‑time processing. The pipeline supports diverse event streams—market data, user behavior, risk control, and transaction logs—enabling use cases like customer churn prevention, risk monitoring, and recommendation algorithms.

Performance Engineering

High‑performance computing challenges are addressed with techniques such as Java Sockets Direct Protocol, hardware‑accelerated storage, and the LMAX Disruptor for lock‑free, single‑threaded processing. The team also discusses the trade‑off between moving computation to data versus moving data to computation nodes.

DevOps and Monitoring

Operational practices include centralized log aggregation, Elasticsearch for search, Kibana for dashboards, and automated alerting. The organization adopts a DevOps mindset inspired by companies like Etsy, emphasizing comprehensive metric collection, end‑to‑end tracing (e.g., Google Dapper concepts), and containerization to support horizontal scaling.

Key Challenges

Three major challenges are identified: 24/7 availability with high concurrency, ever‑increasing transaction volumes and product complexity, and the upcoming need for internationalization across multiple markets with minimal downtime.