Have you seen the management area decoupling architecture? Can help customers solve big problems

Abstract ： What is the decoupling of management area ？ What are the application scenarios ？ This issue will show you .

This article is shared from Huawei cloud community 《【 Hua Wei Yun Stack】【 The big shelf is coming 】 The first 13 period ： Have you seen the management area decoupling architecture ？ Help government and enterprise customers solve big problems 》, author ： Li Junwu Hua Wei Yun Stack Senior architect , Shensi Hua Wei Yun Stack Chief architect of basic services .

Management area decoupling introduction

Similar to the concept of Public Cloud Architecture , Huawei cloud, which is mainly deployed locally by government and enterprise customers Stack Maintain the same architecture as Huawei's public cloud , Include Region、AZ、 The concepts of host groups and other related areas are also basically the same . From cloud platform related components / From the perspective of scope division of cloud services , Usually, the industry will also integrate components / Cloud service deployment locations are divided into Global Scope and Region Scope ,Global Components / A service is a management area belonging to the cloud .

• Global Components / service ： Function in each Region Of , For example, cloud tube 、 Authenticated IAM service 、 Span Region Disaster recovery or business management CSDR、 Span Region Network interconnection CC service （Cloud Connect） etc. , In practical use, it faces multiple under a cloud Region

• Region Scope ：Region Level services are usually used in Region Inside , such as Region1 Of ECS Service can be given Region1 Distribute virtual machines , But it cannot issue another Region2 Virtual machine , Similar to that EVS Block storage 、VPC、EIP、VPN Service etc. .

Here are two special services , It may not be consistent with the reader's understanding , It needs to be clarified .1）VPC-Peering Get through is a Region Inside two VPC Network interconnection , So and CC Different , Also positioned as Region Level cloud services ;2）ELB Although cloud services can add this Region External IP Address as backend , however ELB All you can see is that Region Internal objects and the Region Interoperable objects , So too. Region Class cloud services ;

Hua Wei Yun Stack Management area decoupling of , A light sensitive base is provided to deploy our Global Components and services , This forms a specific deployment area , We call it independence Global District , And you can deploy two independent Global To achieve Global Components / Disaster tolerance of service management .

The original deployment architecture is to Global Components / Service select a business Region Deployment , As the first Region; Choose another one as a backup Region, To achieve Global Components / Disaster recovery of services ; Other businesses are called from Region. Now? , Through Huawei cloud Stack The automatic deployment of the management area realizes the deployment of the decoupling area , In this way, you can expand and change related businesses Region, You can also manage multiple editions Region Change and operate independently of each other . first Global District is the first Region; the second Global Area is prepared Region, To achieve Global Components / Disaster recovery of services ; The rest of the business Region from Region. in other words , It is consistent with the overall master-slave architecture .

chart 1 Open the decoupling design of disaster recovery management area

This is like power supply in some areas , Once upon a time, every household had its own generator ,A The family has been expanded , The generator should also be upgraded at the same time ;A The family generator is damaged ,B Family is A It will waste time and experience for families to continue to supply electricity . But now we use the unified power grid , That is, the management area is decoupled , There is also a backup power grid , If the power grid is damaged , There will be a backup power grid immediately , Reduce time loss ; meanwhile , It is convenient for every family to expand .

The advantages of management area decoupling are also obvious ：

1. Overall, it is lighter and more sensitive ： For deployment / upgrade / Subsequent changes take less time , Evolution is convenient ;

2. The management surface is highly available ： As mentioned above , Independent global The district has the disaster tolerance ability of management , It has good high availability ;

3. Version iteration is more convenient ： Independent Global The zone can manage a large number of different versions Region, Generally follow N-2 principle ;Region Can be expanded independently 、 Upgrades and changes , No other business is required Region coordination , So as to realize various businesses Region Change decoupling between .

4. Deployment is more flexible ： Independent Global Zones can be in independent data centers , No need and jurisdiction Region common DC Deploy

Typical usage scenarios of management area decoupling

The management area decouples and deploys this architecture , It can help government and enterprise customers solve some stubborn diseases accumulated over the years .

One 、 Stock evolution

The overall architecture of the software will increase with the types of cloud services supported 、 The increasing scale of computing nodes is getting larger , At this time , We need to optimize the software architecture . Architecture optimization will bring huge benefits to products , But one of the possible costs , It will make the stock of the original architecture Region Upgrading is more complicated 、 Can't even upgrade , This will lead to different architectures Region It needs to be managed under a cloud . In order to solve the problems of different architectures Region Coexisting problems , Based on management area decoupling , The new architecture can be smoothly evolved and expanded Region.

chart 2 Stock evolution scenario

In the scenario of stock evolution , In addition to automated deployment tools , It also provides Global Components / Automated relocation tool capabilities for services , Except for the same type CPU In addition to the relocation capacity , It also realizes cross CPU Type of data relocation capability . It can realize the distribution of the original stock Region yes X86 Management node of , take Global Components / Service data , Move to ARM CPU Manage the independence of nodes Global Suprazonal , On the contrary, from ARM To X86 Can also support .

Two 、 Cloudy in one

Among government and enterprise customers , Based on company merger 、 Organizational adjustment 、 Personnel changes , Customers will ask for more of the stock HCS Merge into one HCS, In order to achieve a cloud management . Compared with the original cloud Federation , Based on the decoupling of the management area, it provides the ability of multi cloud integration , It can better match customers' management demands of a cloud , To realize the unified management ability of customers .

chart 3 Cloudy scene

First create a new independent Global District , Will need to merge the cloud Global Management area , One by one into the new Global Can then ; The data consolidation here is similar to the above stock evolution Global Components / Data relocation of services , We also provide tool automation capabilities ; such , Through tool automation , You can gradually merge any required multiple clouds into one cloud .

3、 ... and 、 Centralized management

In government and enterprise customers, there is usually a division of organizational headquarters and branches , The headquarters makes decisions and controls the overall trading , Do not deploy business , Branches use cloud resources , Conduct business operations in the territory . Usually , In order to control procurement 、 Divide the organizational model 、 Overall planning of resources and other reasons , Will build a global cloud , The branch's authentication should be applied for at the headquarters . Under the decoupling structure of our management area , By creating an independent without any resource pool and tenant cloud service capabilities Global District , Conduct customer cloud management /IAM Deployment and control of , Branches can be independently responsible and apply for expansion Region Resource pool , Headquarters to audit and approve . In this way, it can meet the cooperation relationship between the headquarters and branches that perform their respective duties on the cloud platform .

chart 4 Centralized management scenario

besides ,Global You can also manage different versions of Region Resource pool , So as to realize the above , Operations between different branches only need to be perceived by the headquarters , Other businesses Region Mutual changes do not affect , It effectively reduces the time of process approval between different divisions 、 Increase of efficiency .

Four 、 Cross pool relocation

stay Region Inside , To support rolling upgrade ,VM Provides thermal migration or cross AZ Ability to migrate ; And many under a cloud Region Resource pool , Due to the expiration of the customer's data center lease 、 Insufficient expansion space of customer stock resource pool and other reasons , Many customers will Region Internal VM And store data to another Region On the demand . Our relocation ability under the decoupling of management area , You can reuse stock data first , take Global Components / Service data moved to Global Independent area , Then expand new businesses Region（ It can be a new architecture business Region）, Realize cross Region Even the relocation capability of different architectures .

chart 5 Cross pool relocation scenario

Cross pool relocation under the decoupling of the management area , You can inherit the existing account of the original customer / password 、 Organization model and related data , Minimize the customer reconstruction operation and use difference perception before and after relocation .

summary

Hua Wei Yun Stack It has been deployed in a joint-stock commercial bank , Decoupling through the management area , Realize centralized management ; In an insurance institution, the stock evolution is completed through this scheme , Protect customers' existing investments ; In a state-owned bank , Realize centralized management and multi region Decoupling independent upgrade . The fact proved that , take HCS Of Global The management area is deployed independently , It can effectively solve customers' concerns about stock evolution 、 Cloudy in one 、 Centralized management 、 Cross pool relocation and other appeals , This architecture , Government and enterprise customers say yes ！

Click to follow , The first time to learn about Huawei's new cloud technology ~