How Ant Group Built a Cloud‑Native, Financial‑Grade Architecture Over 15 Years

Key Points

1. Over the past fifteen years Ant Group reshaped payment and micro‑loan services through technology. The five BASIC technologies—Blockchain, Artificial Intelligence, Security, IoT, and Cloud Computing—remain the foundation of fintech. Computing power is the core capability that must continuously improve.

2. The core of financial transaction technology is distributed middleware, especially distributed databases. High availability, zero data loss, and self‑recovery are essential for financial‑grade systems.

3. Financial‑grade distributed systems are moving toward cloud‑native architectures. Middleware capabilities will be delivered via a service‑mesh layer, and a secure trusted execution environment becomes the baseline; security containers are a strong demand in the industry.

4. The future trend for financial data intelligence is a “Big Data Base”. An open computing architecture with unified storage standards, pluggable engine components, and intelligent SQL is needed to fuse data processing with AI.

5. Diverse computing models—streaming, graph, machine learning—often coexist in business scenarios. Ant Group, together with UC Berkeley, is advancing the next‑generation Ray engine to provide a multi‑modal, fused financial‑grade compute engine.

6. Ant’s open‑source SQLFlow abstracts the end‑to‑end workflow from data to model, making AI applications as simple as writing SQL.

Cloud‑Native Evolution Details

Ant’s distributed middleware has evolved into a cloud‑native stack. During the 2022 Hangzhou Cloud Rex conference, a live demo cut the network cable of a running server in one data center; the system recovered in 25 seconds, demonstrating three‑site, five‑center active‑active resilience.

The company built a technical risk‑as‑a‑service (TRaaS) system that can detect and recover from failures within five minutes, providing proactive fault‑injection through red‑blue testing to continuously improve reliability.

To support cloud‑native transformation, Ant created a dedicated security‑container team and contributed to the Kata Containers project, aiming to provide a trusted, secure container runtime for financial workloads.

Service mesh will gradually replace traditional middleware, sinking capabilities such as distributed transactions, high‑availability, and consistency down to the infrastructure layer, allowing business developers to focus solely on domain logic.

All of these components—service mesh, secure containers, distributed database OceanBase, and risk‑auto‑recovery—form the backbone of Ant’s cloud‑native financial architecture.

OceanBase and Distributed Database

Since 2009 Ant has been developing its own distributed database, OceanBase, which uses the Paxos algorithm to guarantee consistency. Today all Ant financial services run on OceanBase, achieving peak throughput of 42 million transactions per second, clusters with over 1 000 nodes, storage exceeding 2 PB, and single‑table row counts beyond 320 billion.

Open‑Intelligent Computing Architecture

The company advocates an open, pluggable architecture: a unified storage layer with standardized virtual tables, a plug‑in engine layer accessible via SPI, and a unified intelligent SQL layer that serves both developers and data scientists. This reduces development cost, improves efficiency, and enables rapid iteration.

For heterogeneous workloads—real‑time streaming, graph analytics, and machine learning—Ant combines Ray, a multi‑modal compute engine, with a dynamic meta‑computing framework that can execute streaming, graph, and ML tasks on the same data graph.

SQLFlow, an open‑source project, abstracts the full data‑to‑model pipeline, allowing AI applications to be expressed as simple SQL statements.

In summary, Ant’s financial‑grade architecture consists of distributed middleware, a high‑availability distributed database, a technical‑risk prevention system, secure container infrastructure, and a service‑mesh layer, all delivered as cloud‑native services that simplify application development.