当前位置:网站首页>How can cube help kedun's business containerization "in one step" under the demand of cost reduction and efficiency increase?

How can cube help kedun's business containerization "in one step" under the demand of cost reduction and efficiency increase?

2020-11-09 11:32:07 No e45irv7l

Preface

4d30ef64ef6349176b5bdc4623c52190.png


With Docker For the representative container technology shortens enterprise application from development 、 Build to release 、 The whole life cycle of running .Gartner It is speculated that 2022 There will be 75% Global enterprises will use containerized applications in production ( At present, it is about 30%). because Docker It is often difficult to support large-scale containerized deployment independently , So it was born Kubernetes And container layout tools , Solve the organization and management problems of large-scale container .
But in fact ,Kubernetes The system of using is still very complicated , For the development and operation and maintenance personnel of the enterprise , Need to have a certain network 、 Storage 、 System and other technical capabilities . At the same time Kubernetes In the process of cluster deployment , It also needs to face multi node cluster construction and maintenance 、 The choice and configuration of network and storage . The above problems are inevitable in the container deployment and container arrangement of large-scale applications , Kedun faces the same dilemma .

daa70fc155dccf6951bfbd4f924f9824.png


Why does kedun choose containerization ?

dfc5cd6a638b494d61c3b784f5fd728a.png


Shenzhen kedun Technology Co., Ltd ( Hereinafter referred to as kedun ) It is the leading Internet big data service provider in China , It is a national high-tech enterprise focusing on research and application development in the field of cyberspace governance , Committed to network security 、 Netcom 、 The government and large and medium-sized enterprises provide big data Internet public opinion monitoring 、 Network information mining and business monitoring services , So that in time 、 accurate 、 Comprehensively grasp the Internet public opinion information , Quickly resolve the crisis of public opinion . At the same time, we can fully obtain network intelligence clues and market competitive intelligence , Crack down on Internet crimes for government agencies 、 Enterprises improve market competitiveness and provide high-tech edge tools .
The data source of public opinion analysis covers almost all the public information of Internet platforms , Including common information sites and social media 、 We-Media 、 Short video and other platforms . Therefore, we should build a complete and accurate public opinion monitoring system , Not only do we need strong data collection capabilities , You also need to have strong data analysis 、 Value Mining ability , For the bottom IT Architecture presents huge technical challenges .
Before implementing microservices and containerized deployment , Kedun directly deploys the application on the server , Developers spend a lot of time developing 、 Test and production environment configuration , But also to solve the daily emergence of the network 、 journal 、 Monitoring and other issues . With the expansion of the company's business , The whole system is becoming more and more complicated , And it depends on complexity 、 The data is not isolated 、 Logical repetition , So kedun transformed its business into a microservice architecture .
Kedun used SpringCloud Micro service system and Eureka Service registry , The overall business architecture is divided into task scheduling 、 Data collection 、 data storage 、 Data processing and so on , The overall business structure is as follows :

4d30ef64ef6349176b5bdc4623c52190.png


The system collects the data of the registered node according to the task , Schedule effective acquisition targets to data acquisition nodes , The data acquisition node processes the target according to the algorithm , Transfer data results according to their content type , Transfer to based on MySQL、MongoDB、HBase、Elasticsearch、 Storage queues for services such as object storage , The data is cleaned by the processing chain 、 Resource file split download 、 Model to predict 、 Business tag of each subsystem 、 Data push service, etc .
In the microservice phase , As the number of applications increases , One release often involves multiple applications , This puts forward higher requirements for the team's automatic operation and maintenance level . therefore , The team started to container the application . Each module is packaged into a mirror image , Can run on any platform , Easy to achieve business migration and expansion ; There is no need to repeatedly configure the environment , coordination Gitlab It can be very convenient for continuous delivery and deployment , You can also isolate applications .


Cube Faster and lighter than the container

4d30ef64ef6349176b5bdc4623c52190.png


The use of containers does solve the delivery efficiency faced by kedun before 、 Operation and maintenance cost and environmental consistency , But self built Docker There are still some problems with the service , for example :
  • Need to buy fixed resource specification host , The cost input is relatively high ;

  • It can only be mounted through the host , Cumbersome operation ;

  • Only support single IP, Binding extra IP It's very complicated ;

  • In case of failure, additional control and scheduling system should be installed to restart ;

  • Only through namespace and cgroup Weak isolation ;

  • Need to use docker Command to create , High learning cost .


adopt UCloud Container instance Cube, Users only need to provide a packaged container image , In seconds , Realize batch deployment of containerized applications , You don't need to buy a host or K8S colony , And only pay for the resources consumed by the container's actual operation .
Besides Cube It also has the following advantages :
  • Use... On the Internet UCloud VPC The network provides intranet services , And UCloud Other cloud products get through ;

  • The storage is directly attached to the cloud disk , High reading and writing performance , Easy to operate ;

  • Cube The control and scheduling system automatically restarts the container , Strong self-healing ability ;

  • UCloud Massive resources support , Large clusters avoid single node failures ;

  • Use Firecracker Virtualization technology realizes strong isolation at the virtual machine level .


be based on Cube The advantages of , Kedun has migrated its data acquisition node to Cube, The log collection scheme is accessed by the original host local mapping , Send to instead Kafka queue . Due to the nature of containerized applications , Kedun doesn't need to transform its business , The original image can be directly used in Cube Deployment application , Achieve smooth migration . and Kafka Application , It also improves log collection 、 The efficiency of Management .
Cube Help kedun to achieve rapid business roll back and horizontal expansion . follow-up , Kedun plans to introduce K8S colony , And migrate more services on the data processing chain to Cube.


summary

4d30ef64ef6349176b5bdc4623c52190.png


Kedun's business architecture has changed from traditional server deployment mode to microservice architecture 、 Containerization , And then to the choice UCloud Container instance service Cube Migration , It's enterprises that reduce resource costs 、 The way to improve the efficiency of operation and maintenance of development , It also fits the context of the rise of cloud native technology . Whether it's a container Docker、Kubernetes still Serverless Architecturally Cube, Finally, it will return to the core value of reducing cost and increasing efficiency for enterprises .















版权声明
本文为[No e45irv7l]所创,转载请带上原文链接,感谢