current position：Home>Financial IT architecture - Analysis of cloud native architecture of digital bank
Financial IT architecture - Analysis of cloud native architecture of digital bank
2022-05-15 07:20:42【Young and promising 2025】
I've spent almost... On and off recently 2 To 3 I finished reading the book published by electronic industry press in a week's time 《 Financial grade IT framework ： Decryption of cloud native architecture of Digital Banking 》 This book , The author is the technical editorial board of e-commerce bank .
This book is in line with the current trend of distributed architecture transformation in the banking industry , The content is the leading technology tested by practice, and introduces the e-commerce bank IT Technical architecture evolution route , It covers distributed 、 unitized 、 Elastic hybrid cloud 、 Cloud native multiple infrastructure domains , At the same time, the technical risks are introduced 、 Safe and trustworthy 、 Business architecture and other technical practical experience .
For this book, I actually searched for cloud native related books very early , But at first glance, the title is finance IT The of architecture has not attracted much attention , Later, I was lucky to read the book after being donated by Huazhang , After reading, I have a feeling that I hate to meet you late .
I'm not familiar with e-commerce banks , I began to wonder why e-commerce banks were able to write the book , After a search on the Internet, I learned that e-commerce bank is the first commercial bank in China whose core system is based on cloud computing architecture initiated and established by ant group as a major shareholder . It was approved by the CBRC as one of China's first 5 One of the private Banks , On 2015 year 6 month 25 Officially opened on . Therefore, the business volume of e-commerce banking services for small and micro enterprises is also large , At the same time, because ant group is the major shareholder , So in the whole IT Ali is adopted in the construction of infrastructure and technical architecture , Ant financial has a lot of open source technology components and architecture . In other words, the technicians of e-commerce bank are the former technical experts of Alibaba .
Therefore, it is not surprising that the technical editorial board of e-commerce bank can write the book .
But why the whole editorial board, not a single author , This part covers a wide range of contents , from IT Infrastructure , To cloud native , To be safe and reliable , Wait at the middle stage . Each individual chapter has enough depth , It's hard for one person to master so many subdivided technical fields .
And basically the content of each part comes from IT Summary of front-line practice of architecture construction , The sharing of many experience points also comes from a lot of practice , This is not the point that ordinary technical books can summarize .
According to the content introduction of the book, we can also see the same content description .
This paper introduces the history of e-commerce bank since its establishment IT Technical architecture evolution route , It covers distributed 、 unitized 、 Elastic hybrid cloud 、 Cloud native multiple infrastructure domains , At the same time, the technical risks are introduced 、 Safe and trustworthy 、 Business architecture and other technical practical experience , We hope to share with readers the financial level of e-commerce banks IT A unique exploration in technology , Talk to you about the financial level in the digital age IT The development direction of Architecture .
The author of this book is the core architect of e-commerce bank , Deeply participated in the whole process of relevant technical schemes from preliminary design to later production , The content is clear and authoritative . This book focuses on the technical practice of e-commerce banks , The content described represents the leading technical direction , Relevant technologies have been tempered by the real production environment , It includes the unique practical experience of the technical team of e-commerce bank , The core technology described in the book won the award of... Issued by the people's Bank of China “ Bank Science and Technology Development Award ” The second prize .
This book is not just for the financial industry IT Reading for employees , It is also applicable to large enterprises that need to build a group IT Infrastructure and cloud native architecture . Many financial industries are in the process of distribution , High availability , Elastic expansion , The accumulation of architecture and practical experience in security is fully applicable to the digital and cloud native architecture transformation of large enterprises .
Here are some key points of this book , Some of this book PPT The pictures come from the public technology sharing of Jiang Yimin, one of the authors of this book .
The evolution process of the Three-Generation architecture of e-commerce bank
From this picture , E-commerce bank has mainly experienced the evolution of three-generation architecture .
The first stage is mainly based on cloud platform + Built on a distributed architecture . The whole deployment mode is the dual active mode in the same city . here we are 2017-2018 The second stage of the year , On the basis of double living in the same city, we need to build a remote data center , I hope this remote data center can carry business traffic in the daily process , Can provide external services with Hangzhou data center at the same time . While meeting the requirements of remote disaster recovery, consider improving the whole system IT Asset utilization of infrastructure , Therefore, we have created a modular and flexible architecture , It is a deployment structure of three places and five centers . from 2019 year , E-commerce banks began to pay attention to the cloud native architecture , Including the introduction of some products , Design and build relevant capacity . In the process , We have also built a hybrid cloud elastic Architecture , It has the ability to schedule resources between two clouds .
Briefly summarize the whole evolution route, and the key technical points are ：
Data split vertically , Data horizontal split , Distributed architecture construction , Cloud computing platform construction , Modular multi active architecture , Flexible architecture , Elastic architecture construction , Hybrid cloud architecture construction , Native Cloud Architecture .
An important concept is mentioned in this book , namely Cellular architecture .
Each unit is one from the traffic layer , Application layer to data layer integrity , autonomous , Independent ecosystems , It can provide most services for users , Data access shall be completed in a separate unit as far as possible . Therefore, a unit can be deployed to any region , At the same time, the unit and the unit can back up each other .
When you see this, you may contact distributed and micro Services .
Because when I talk about micro services, I often emphasize that the focus is the separation of monomer to micro Services , Every microservice is implemented from the data layer , Logic layer to application layer , From demand , Design , Develop build , Compile and deploy independent autonomy . But these are all talking about modularization at the software level .
The goal of modular architecture is Implement software level + Common unitization at the hardware level . It's a bit similar to the concept of all-in-one machine that was popular many years ago . The integration of software and hardware makes it more convenient for each unit to cross region , Multi data center migration capability , It also further enhances IT High availability and redundancy at the infrastructure level .
From elastic computing to cloud native
In tradition PaaS Elastic computing and dynamic resource scheduling are also discussed in the platform , But tradition PaaS It's hard to be completely flexible , automation , Fully auto scalable elastic architecture capability .
On the one hand, flexibility is automation , On the one hand, it is necessary to realize both scalability and contractibility .
Since the birth of e-commerce bank ,IT The system is built on the private cloud ,IaaS Based on Alibaba cloud's proprietary cloud base ,PaaS Financial cloud construction based on ant group , Naturally, it has distributed computing power . And have financial level security , Highly reliable , High availability .
Cloud native technology includes micro Services , Containerization , Immutable infrastructure , declarative API, Service grid, etc . Cloud native architecture is a set of architecture principles and design patterns based on cloud native technology , The key point is to separate some non functional code in business processing logic , So that cloud facilities can take over a large number of non functional features in the application （ Such as elasticity 、 toughness 、 Security 、 Observability 、 Gray scale, etc ）.
You can see from the picture above that , Any code logic in traditional architecture includes business code , Non functional code and third-party dependencies . And in the cloud native architecture , The hope is to separate business and Technology , Business function developers only develop business function codes , Platform developers are responsible for the realization of various non functional requirements and observability .
A lot of non functional features , Including elasticity 、 Capacity 、 Safety observability 、 Grayscale, etc. will gradually sink to the infrastructure of the bottom layer , Especially high availability 、 Disaster resilience 、 Capacity guarantee 、 Safety features , There are also some features that can be operated and maintained , Gradually let the infrastructure take over . In this case, it can be seen that , Some changes that will happen in our deployment . See from the bottom right corner of the figure , The container will be further split , Split into a process , Side car （sidecar） A process .
After reading this book, another biggest feeling for me is .
In tradition IT Architecture transformation , Micro service , And the development of cloud native technology . In order to achieve flexibility and flexible expansion , Implement distributed , To achieve decentralization ,ServiceMesh Service grid application is imperative .
Especially through containers, clouds and Kurbernetes The scheduling and arrangement of container resources are realized , After immutable infrastructure , It is a dynamic side car agent , take Sidecar And business containers are packaged into one POD It is possible to manage . This has also greatly promoted Mesh The development of grid technology .
After thorough meshing, you will see , In addition to the north-south traffic, it also needs the upper load balancing equipment to solve it , Other problems can be solved through side car agent and decentralization , Security of interface services and traffic , journal , Various governance capabilities such as current limiting and fusing .
E-commerce banks also experienced rich banking in the early stage , This is heavier , It contains all the dependencies required for application publishing and deployment , Not limited to some key RPC、 news 、 Database etc. SDK. The smallest deployment unit is also a container , Further division in the primary cloud .
Containers are divided into APP The container of , Follow Sidecar The container of . According to current practice , Mainly including Service Mesh Inside MOSN The container of , also DBMesh The container of . These two containers solve RPC、 news , There are also forwarding at the database and cache levels .
One of the biggest benefits of this model , Namely MOSN Follow DBMesh It can evolve independently , That is, you don't need to cooperate with the upper business container , Complete some upgrades and releases by yourself .
stay Mesh In the standardized Architecture , The actual Internet speed bank is divided into two different Sidecar, One is to solve the problem of east-west traffic and service governance MOSN, The other is DBMesh agent .
DBMesh Agency is a very important idea .
Simply put, it is distributed in the database , Or after the database is split horizontally and vertically , The traditional logical idea of architecture is to add one on each unit DaaS Unified data access layer . however DaaS The deployment of the layer itself is centralized , That is, all the traffic of the underlying database must go through DaaS Layer of APP The server , that DaaS The layer itself becomes a centralized node .
Therefore, the best solution should be DaaS The ability of the layer to act as a separate container Mesh To all POD In the group .
Flow isolation and flow allocation
The core value of cloud native architecture is to realize fine isolation of traffic .
Thoroughly in the whole architecture Mesh After transformation , You can see in the Sidecar The side car agency can handle the routing of traffic well , Flow isolation and fine control capability .
Based on new cloud native capabilities , In the process of traffic forwarding , Can flow through MOSN Mark when the side car agent , Let it route to some specified containers , You can do under different business requests , It will be routed to different container clusters , The interaction between businesses is further reduced .
The most typical is the hot account problem encountered in production , It can easily lead to the transaction jitter of the whole bank . If we can identify the hot spots caused by different businesses , Can achieve effective isolation , The occurrence of hot spots will not affect other business scenarios that do not generate hot spots .
Under the new cloud native architecture , be based on mosn It can create more fine-grained traffic allocation , From the data center level to a single application level . You can find some insensitive application services to cut the flow first , Avoid affecting key business content .
In the absence of Mesh Before melting , If only through load balancing devices or gateways, such fine traffic isolation is often difficult to achieve , This is also Mesh Another key capability brought about by Urbanization .
From full link voltage measurement to chaotic engineering
For the complexity of cloud native architecture , It is an inevitable trend to introduce chaos into engineering .
2018 year , Chaos Engineering （Chaos Engineering） Become CNCF A new field of Technology .
For the chaotic engineering part CNCF The foundation included it in the observability section . stay 2020 year 8 The observability technology survey was completed in June , Members of the end-user community were asked about their assessment of 、 What observable solutions were tested and subsequently adopted . Yes 283 Sort and review data points , Determine final location .
Some key points of chaos Engineering .
One is that chaos engineering is not just done in the test environment , But directly simulate in the production environment . In other words, the test environment is difficult to fully simulate the production environment , Then it is necessary to conduct node fault simulation in the production environment , Also confirm the whole IT The stability of the architecture in the production environment .
The second is the real integration of business concurrency performance test and reliability test . In traditional testing, the two are often separated , It's hard to fully simulate... In a test environment .
The third is the uncertainty of fault simulation itself , The first is the uncertainty and randomness caused by the fault itself , The second is the uncertainty of the combination of various fault scenarios .
The complexity of distributed architecture under the cloud , It is an inevitable trend to introduce chaos into engineering , Just like when I talked about cloud native and micro services earlier , introduce ServiceMesh Microservice governance is also an inevitable trend .
At present, chaos engineering is a booming technical field , And it's getting more and more attention . It is also a set of practical measures proposed to deal with the complexity of distributed architecture , Rigorous engineering practice principles , Methods and tools . Chaos engineering is based on the idea of anti fragility , Simulating a fault is just a means , The core goal is still to improve the stability and observability of the system , Identify risks early , And optimize and solve .
Any production business system , Should not be a problem after a failure , It should be active discovery , Risk driven mechanism of active defense . This is the great value and role of chaos engineering in the cloud native architecture .
This book gives a brief introduction to the full link voltage measurement and its methods and actual combat , However, from full link voltage measurement to chaotic engineering methodology , It is an inevitable change to deal with the complexity of cloud native architecture .
For cloud native architecture , In fact, in the original article, I also talked about several key technical points, and the ability is worthy of in-depth study , It mainly includes the following contents .
- Chaos engineering and observability
- ServiceMesh And decentralized service governance
- Highly flexible and automated elastic extension Architecture
- Distributed middleware and distributed transactions
- Integrated R & D, operation and maintenance platform and DevOps
- Traffic management
- Secure and trusted architecture
The above content will be the development of cloud native architecture and technology , And gradually become mature .
Of course, this book also has some shortcomings , Because it is in multiple technical teams , Many people's works , I feel that the logical relationship between the contents described in each chapter is not rigorous , At the same time, the same content will be described in different chapters . This is a common problem when many people work together .
But there is still a difference between the next book and simple patchwork , One flaw cannot obscure the splendor of the jade , The overall framework content is still complete , Many practical contents and experience sharing are worthy of careful study and reference . Recommend the book again . In particular, large and medium-sized enterprises are facing digital transformation , Cloud native IT Technical directors and architects read .
Finally, dry goods , An enterprise class developed a few years ago saas The system is all open source , Interested partners can study .
weiit-saas Is a Java Open source project , Belong to weiit The team develops its own products , Intended to be packaged through technology , Let the enterprise do not need code development , Help enterprises generate small programs with one click 、 official account , Let the enterprise have the independent brand self operated Mall . Product competitors 《 I like it 》、《 Micro Alliance 》.
author[Young and promising 2025],Please bring the original link to reprint, thank you.
The sidebar is recommended
- [opencv] learning materials
- Third, Xin'an has not joined the Euler open source community, and all password card products have completed the openeuler compatibility certification
- SA login failed, vs2010sql server2880r2
- Non recursive implementation of hierarchical traversal and preorder traversal of the tree (stored in child notation in pointer mode)
- Stone merging (ring)
- Vs2022 reference of external dependency Library
- Candy (multiple of K required)
- Stone game
- Calculate string distance
guess what you like
Acwing daily question 2022 / 5 / 10 [3746. Niu's academic circle II]
The adjustment of plane network and elevation network is obtained by using Keshan software
The practice of native Flink on kubernetes in Little Red Book
Multilingual mixed programming
Flink CDC meetup · online, 5.21
Why is the real-time data warehouse class database not equivalent to HTAP database?
Vs code plug-in recommendation
Design of Web rights management (1) -- Analysis of key points of design (3)
P4551 longest XOR path
Chain forward star template
- Redis -- realize the like function
- What are the problems with NoSQL?
- What aspects does the class database of real-time data warehouse include?
- [MQ] achieve mq-08- configuration optimization from scratch fluent
- Rongyun x dsport: win the first drop of blood of game social "real-time community"
- Druid source code reading 5 -- discuss why Druid doesn't use atomiclong
- Construction practice camp - graduation summary
- [module] environment variable management tool module installation and use
- Kernel heap bypass SMEP, SMAP & modprobe_ path
- Getting started with mqtt: online debugging and connecting to alicloud
- Leetcode deletes the penultimate node of the linked list
- Design of Web rights management (1) -- Analysis of key points of design (4)
- Tencent cloud database growth index ranked first
- Three ways! Go error handling best practices
- Live broadcast at 20:00 tomorrow night ｜ open the box Tencent cloud: play lighthouse, novice friendly and light-weight cloud!
- New study conclusion: sleeping more than 2 hours on weekends reduces the risk of depression by 48%
- Retinal vessel segmentation based on u-net model
- Applet X-axis fade in animation component
- Explanation of wechat applet interface call (accesstoken, applet code login)
- Single cell column - how to give orig Ident, change your name
- Fonts best practices
- Wonderful express ｜ April issue of Tencent cloud database
- Illustration: what is the difference between layer 2 and layer 3 switches?
- Activity Notice | timing adjustment of 2022 deterministic network technology and Innovation Summit
- In order to seize the capacity of 5nm chips, AMD will pay an advance payment of US $6.5 billion to TSMC, grofangde and other suppliers; Germany will adopt stricter antitrust rules for Google meta
- It is reported that TSMC will promote the 1.4 nm process next month; Taobaoyuan universe trademark rejected
- Online binary 8-hexadecimal conversion tool
- [paper notes] epsanet: an efficient pyramid sequence attention block on revolutionary neural network
- IndexError: shape mismatch: indexing tensors could not be broadcast together with shapes , 
- What are the development stages of time series database in recent years?
- What are the shortcomings of the data model processed in the first stage of time series database?
- What are the shortcomings of the data model processed in the second stage of time series database?
- What are the development trends of time series database?
- What are the characteristics of cloud native multimode database lindorm?
- What are the functions of cloud native multimode database lindorm?
- Variance, standard deviation, mathematical expectation
- Two dimensional Gaussian distribution
- Collaborative process and channels (CSP: kotlin, golang)
- SQLite3 custom function (UDF)