Cloud-native architecture transformation for digital finance
In the era of digital finance, the industry widely agrees that cloud-native architecture is a key trend for financial IT. However, transforming an IT architecture does not happen overnight. While exploring and applying new technologies such as cloud-native, it is essential to integrate them with traditional models. This innovation must follow a stable and practical path to ensure that the architectural transformation can reliably support and even lead business innovation.
Through Ant Group's internal architectural changes and by making our financial technology available to others, we have explored and defined a path for implementing a financial-grade cloud-native architecture. This path meets the demand from financial institutions for distributed transformation and stable, innovative IT upgrades. This document is a starting point. We aim to share Ant Group's cloud-native practices and experience and collaborate with the industry to advance architectural transformation and digital finance innovation.
What is cloud-native?
Industry leaders and startups share common architectural traits. They focus on mobile to enhance the user experience, innovate quickly, ensure continuous service, and use scalable infrastructure. These are also the goals for the digital transformation of traditional institutions, such as those in finance. A cloud-native architecture is designed to deliver this value.
The term cloud-native first appeared in the tech community in 2013. It represents a set of advanced architectural concepts, including microservices, agile development, DevOps, continuous integration and deployment, containers, reliability, high elasticity, and scalability. Subsequently, many excellent projects emerged from the community. The Cloud Native Computing Foundation (CNCF), part of the Linux Foundation, was created to unify these standards.
In June 2018, the CNCF released version 1.0 of its cloud-native definition: "Cloud-native technologies help organizations build and run scalable applications in modern, dynamic environments such as public, private, and hybrid clouds. These technologies include containers, service meshes, microservices, immutable infrastructure, and declarative APIs."
The enterprise journey to cloud-native architecture
To understand cloud-native and the adoption path for traditional enterprises, we will first review the evolution of cloud computing and enterprise architecture. Broadly speaking, the journey from a traditional architecture to a cloud computing architecture has three stages:
1. Cloud-based: This is the first step of migrating data centers to the cloud, which can be a public cloud or Apsara Stack. Virtualization improves resource utilization and operations and maintenance (O&M) efficiency. However, middleware and the underlying platform still require significant attention. Factors such as high availability, scaling, and deployment must be considered for each component.
2. Cloud-ready: The decentralized, service-oriented transformation is mostly complete. Applications are stateless and managed separately from persistence layers such as caches and databases. As application services scale out, mature DevOps and automation platforms are needed for unified management.
3. Cloud-native: Practices such as microservice architecture, containers, orchestration, and DevOps accelerate application delivery and deployment. Platform capabilities are abstracted and standardized. Combined with microservice administration, agile resource orchestration, and control technologies, developers can shift their focus to concentrate more on business logic, achieving "Code Defined Infrastructure/Everything".
Two key trends emerge from these three stages of evolution:
Trend 1: The focus of architecture gradually shifts upward during application migration to the cloud
As development and O&M concepts evolve and tool platforms mature, the capabilities of underlying platforms and middleware are abstracted. They are decoupled from the business applications that run on them. This lets software teams focus more on developing business logic. Platform technology and business development have clearer boundaries and collaboration mechanisms. Business development teams do not need to worry as much about non-functional features such as disaster recovery and scaling. They can concentrate on implementing business logic.
Trend 2: As container and orchestration technologies mature and unify, the technical direction of the PaaS layer stabilizes
The journey of migrating enterprise applications to the cloud reflects the industry's growing understanding and the maturity of cloud computing technology. Business applications have shifted from relying on the Infrastructure as a Service (IaaS) layer to the Platform as a Service (PaaS) layer. These PaaS platforms solve many real-world problems and reduce reliance on underlying infrastructure. However, they also created a problem. PaaS products can be intrusive to application development and O&M. For vendors, this creates customer loyalty. For users, it can mean vendor lock-in. The industry only began to move in a unified direction after container orchestration technologies such as Kubernetes matured and gained widespread community and user acceptance.
Cloud-native may be a 'silver bullet', but architectural transformation does not happen overnight
Based on discussions in the tech community, technologies such as cloud-native, microservices, DevOps, agile, and containerization seem to be the future of IT infrastructure. The industry widely believes that a cloud-native architecture accelerates delivery, lowers operational costs, supports elastic scaling, and ensures business continuity. This allows organizations to adopt new technologies and expand their reach. From a basic perspective, a cloud-native transformation can seem as simple as putting an elephant in a refrigerator:
Microservices: Break the elephant into many small elephants.
Containerization: Open the refrigerator and put the small elephants inside.
Dynamic orchestration: Arrange the refrigerators to optimize space and resource use.
Cloud-native is an excellent direction for architectural transformation, especially for enterprises such as banks. However, applying these new technologies requires careful consideration of all factors. This requires a complete, proven set of methods and a practical implementation path.
In reality, transforming the architecture of a financial system cannot happen overnight. Integrating new technologies with traditional architectures presents even greater challenges. For example, banks have been building systems for everything from electronic and online banking in the 1980s and 1990s to modern mobile banking. New business demands constantly create opportunities for architectural change. Banks have mature IT systems and recognize the value of cloud-native and distributed architectures for their business and IT delivery. However, these mature systems carry a heavy legacy burden from many critical legacy systems, which presents significant challenges for architectural transformation.
Challenge 1: How to ensure a stable transformation for financial IT architecture?
The word 'architecture' comes from the building industry and carries a metaphor of construction. From the blueprint to the foundation to the final structure, builders try to avoid changes because changing a building's core structure or foundation is very expensive. The software industry borrowed this concept. We use the term 'architecture' to describe the set of design decisions that are difficult and costly to change. Traditional IT architecture and early software development used the waterfall development model. This model involved extensive top-down design before construction began. Once the design was final, it was difficult to change. A single design change could have a ripple effect across the entire system.
Challenge 2: How to balance architecture in demanding financial scenarios?
Choosing an architecture for a financial IT system requires great care. For a bank's core system, this choice is a strategic decision with long-term effects. Today, many banks are considering a gradual upgrade from monolithic to distributed architectures. Delivering an IT architecture that provides business value is about more than just choosing a trendy technology. Selection criteria include service-oriented architecture, agile delivery, open API platforms, and total cost. Feasibility and ecosystem maturity are also important. Therefore, it is also important to consider the long-term vision and roadmap of the technology product. For example, does it have a public cloud strategy? Does it provide a complete migration path from a classic to a modern architecture? Are there real-world case studies and support systems in the industry?
Challenge 3: How to address the key technical challenges of core financial systems?
The growth of inclusive finance and Internet businesses has made the ability to handle large-scale, high-concurrency transactions a standard requirement for the business systems of financial institutions such as banks.
The technical challenges for core financial systems differ from those of other large-scale internet systems. During digital transformation, technology decision-makers in financial institutions must plan for a reliable infrastructure and an elastic application data architecture. This planning ensures business continuity and data security under varying business loads. Financial-grade transaction systems also have stricter requirements for data consistency and the cost of transactions. Managing the risks and security of customer funds is also more complex. As the speed of product iteration increases, mature architectures, processes, and risk-control mechanisms are also necessary. These mechanisms ensure a seamless user experience during application and system updates and maintenance, even under extreme business pressure.
Building a financial-grade cloud-native distributed architecture together
Based on more than a decade of experience gained from Ant Group's evolution toward a cloud-native architecture, we have summarized the following strategies to address the challenges that financial-grade IT systems face during their cloud adoption journey:
The transformation from a classic IT architecture to a financial-grade cloud-native architecture should be based on business value. It should use incremental delivery, stable innovation, and evolutionary planning and build middle platform capabilities to respond quickly to market demands.
When selecting an architecture and technology, it is important to find the most balanced path to meet business development needs and handle demanding scenarios, rather than focusing only on features.
A core financial system based on a cloud-native architecture must also address several key issues, including data-center-level scalability, regional disaster recovery, and distributed transactions under high concurrency. It must also use flexible resource scheduling to optimize costs.
Ant Group will continue to share its experience and technology with the industry, providing practical paths and insights for architectural transformation. We also hope to collaborate with the tech community and industry partners to jointly build financial-grade cloud-native standards and best practices. This allows business applications to focus on agile delivery while being natively equipped with financial-grade infrastructure capabilities, including high availability, consistency, massive concurrency, and elastic scaling.
We call this financial-grade, cloud-native distributed architecture solution SOFAStack™ (Scalable Open Financial Architecture Stack). It is the result of over a decade of developing a financial-grade distributed middleware system within Ant Group and incorporates the architectural practices from Alipay's critical financial transaction systems. All SOFAStack products and technologies have been proven in Ant Group's own demanding financial scenarios. They support agile business iteration while ensuring the security of financial transactions. They also meet the needs for off-site disaster recovery and low-cost, rapid scale-out. SOFAStack addresses the challenges of transforming traditional centralized architectures and supports financial innovation by enabling large-scale, high-availability distributed system architectures.
Additionally, components within the SOFAStack ecosystem are gradually being open-sourced to the community at https://github.com/alipay. We hope to collaborate with the tech community to improve key products. This collaboration will help organizations implement a financial-grade cloud-native architecture with more agility and stability.