preface

On the weekend, I was fortunate to attend the summit “In a Word, ten years of Architecture into a Sentence” organized by “The Class”. This conference invited ctos and architects from mainstream Internet companies in China to share software architecture technology, covering multiple technical aspects such as architecture, database, DevOPS and machine learning. Jingdong, 58, Yihaodian, Yonyou and other companies came to share. As can be seen from the companies present, there is no Google, Facebook, or BAT, so there is a certain distance from the first-tier level. But because of this reason, what is shared may have more reference value for the company in the initial stage. This time, what is shared is more realistic, and the core essence of the organizer is well captured. A. The activity organization is also quite good, can make fun of the place is that the remote control is too low, the speaker volume is too loud, the chair is too hard, PPT is too ugly…… Ten thousand words are omitted here. Okay, that’s not the point. Let’s take a look at the poster, which is the one below.



The pictures in this article are from the scene, please forgive me.

The main content

An overview of

The summit started at 9 PM and ended at 6 PM with a one-hour break. There were eight sharing sessions. Here are eight shared themes. – Ten year architecture technology changes, Talk about coping with high complexity of architecture – micro-service architecture practice – EVOLUTION of JINGdong Open Platform architecture – Evolution of IM architecture – application (business) architecture – balance and bridge between architecture and technology – cluster solutions and implementation of deep learning – Devops in elastic cloud platform How to structure diversified product form – database sub – database sub – table middleware practice below I picked a few more general and dry goods to share a simple introduction.

Ten years of Architecture technology Change, on the high complexity of architecture

This share is made by CTO of Bigeye Technology, which discusses the nature of complexity of software system, discusses the causes of complexity, summarizes the changes of software architecture technology, and analyzes the countermeasures to reduce complexity of software architecture from multiple dimensions. From the quality of code, encapsulation and decoupling, quality to project management and team culture to discuss the software system complex reasons and countermeasures, and combined with their own practice. Emphasis is placed on the architecture level of C4

  • System Context – Boundary definition
  • Container – Service definition
  • Component – Decoupled
  • Class – Concrete implementation

Micro-service architecture combat — JINGdong open platform architecture evolution

This paper introduces the evolution process of the technical architecture of JD open Platform from 2010 to 2016. The main process is as follows – 1.0, a large system (all service concentration, minicompany +Oracle) – 2.0, using the form of MQ to realize asynchronous, using the form of sub-database and sub-table to achieve read and write separation, stand-alone room – 3.0, Docker, self-developed Redis, mysql, At present: Nginx + Lua + HaProxy + level 1 cache + service + JVM cache + multi-level live + self-developed MQ+ sub-database sub-table mysql+ traffic limiting degrade fuse

Devops landing on an elastic cloud platform

This paper describes the architecture of Yonyou’s iUAP platform, which is a development operation and maintenance platform based on the underlying hybrid cloud built by Alibaba, Huawei and Liberty cloud platform, combined with Docker and other technologies.

The system process



Devops development architecture



The system is mainly built based on Docker technology, with high availability and high scalability. The underlying cloud platform adopts hybrid cloud composed of multiple cloud platform providers.

At the same time, Jenkins – based CI and CD systems are developed.

Some of the potholes that Devops has trodden (thanks for sharing)



The core of the enterprise

1. Can be monitored

2. Team culture

3. Organizational form

Deep learning cluster solutions and implementation

In my opinion, this is the most valuable share of this activity, which combines artificial intelligence, big data processing, DevOPS and other hot topics. And provide quite specific solutions.

Go directly to the architecture diagram of the solution

The characteristics of the solution, users’ feedback as a driving force for system improvement to form continuous improvement closed-loop, through Kubernetes docker cluster management, realize the dynamic intelligent expansion of the training platform.

conclusion

There’s one thing you can all agree on:

There is no best architecture, only the right one for the business

In this era, pure software and IT technology are no longer the source of the core competitiveness of most companies, and business is often more important, but the most important is indeed the company or corporate culture. There are several commonalities:

  • Docker is a good thing, which has been mature and verified by JINGdong’s scale

  • Devops is an inevitable trend in companies that make products

  • Agility has become a basic requirement for most software companies

  • Redis is a good thing. It’s a must for resistance

  • There are a few things high availability and high concurrency architecture needs: MQ, Redis, Docker, Haproxy, ElasticSearch

    The final conclusion is that I am more convinced that we are on the right track.