Alibaba cloud MSE supports go language traffic protection

author ： Bobby

summary

With  Go  Language 、 Widespread adoption of cloud primitives ,Go  Language is more and more widely used in micro service scenarios , Yes  Go  Governance of language microservices 、 The demand for current limiting and degradation is also getting stronger .

stay  Go  In language , Although the community provides  go.uber.org/ratelimit  Equal current limiting reservoir , But one of them , Insufficient multilingual support , Only support  Go, second , The function , Current limiting degradation will be subdivided into flow control 、 Isolation 、 Fuse 、 Hot spots and other functions , Dynamic configuration is not supported , Not perfect in function support .

Alibaba cloud micro service engine （MSE） combination  Sentinel, to  Go  Language 、Java  Language applications bring microservice governance capabilities . This article will introduce you how to  Go  Use current limiting and degrading capabilities in language microservice applications .

What is current limiting degradation

In microservice applications , Current limiting degradation is mainly divided into three steps ：

• Target:  For what kind of traffic

• Strategy:  Current limiting and degrading strategy

• FallbackAction:  Triggered behavior

such as , Create an interface for the order （Target）, We limit the request to  1000QPS（Strategy）, After triggering current limiting , The request returned an exception （FallbackAction）.

In Ali cloud  MSE  Support through open source  Sentinel  To define resources , And from  MSE  obtain 、 Apply current limiting degradation rules , The overall access is as follows ：

How to use  MSE  Current limiting the drop

Application access

• download
   **MSE Go SDK 
  [
   
  1]
   **, Unzip to the of the project ./pkg-custom/github.com/aliyun/aliyun-mse-go-sdk-v1.0.7  Catalog .

• You can refer to  SDK  Of  example  Examples in the directory , To access .

• stay  go.mod  In file , Add the following dependency declaration ：

require (
 github.com/aliyun/aliyun-mse-go-sdk v1.0.7
)

replace github.com/aliyun/aliyun-mse-go-sdk => ./pkg-custom/github.com/aliyun/aliyun-mse-go-sdk


• During application startup , add to  MSE SDK  Initialization command ：

import (
 mse_sdk "github.com/aliyun/aliyun-mse-go-sdk"
)
//  Add the following code to the initialization logic of the application .
// Sentinel core Initialization of is included here . If you have called Sentinel Initialization function for , Need to get rid of it. .
err := mse_sdk.InitMseDefault
if err != nil {
 log.Fatalf("Failed to init MSE: %+v", err)
}


Application deployment

In cloud native deployment , The usual deployment specification is  12  factor  

[

 

2]

 , About configuration ,

It is recommended to store the configuration of the application in the environment variable . This application is deployed , Just switch between different environment variables , You can access different environments .

alike ,MSE-Go-SDK  It is also recommended that you access through environment variables  MSE, The environment variables used are as follows ：

After connecting according to the above steps , You can see the applications we access on the application list page ：

Resource definition

At the beginning of the article , We mentioned current limiting degradation =Target+Strategy+FallbackAction. So the first step is to define  Target.

• Sentinel  Define resources

MSE  Support user access  Sentinel  Defined resources , Just wrap the business logic with the following code block ：

import (
 sentinel "github.com/alibaba/sentinel-golang/api"
)

// Entry  The method is used for burying points
e, b := sentinel.Entry("your-resource-name", sentinel.WithTrafficType(base.Inbound))
if b != nil {
 //  Request is flow controlled , It can be downloaded from  BlockError  For current limiting details
} else {
 //  Requests can be made through , Write your business logic here
 //  Make sure that after the business logic is over  Exit
 e.Exit()
}


Of course , In daily business development , Developers often provide services through microservice frameworks , such as  dubbo-go、Gin、gRPC  etc. , spontaneous , We hope to register these services as resources ：

• dubbo-go  How to define resources

Just import  dubbo-go adaptor  You can automatically register resources to  MSE：

import (
 _ "github.com/alibaba/sentinel-golang/adapter/dubbo"
)


• Dubbo  Application is in code through  import  In the form of package  Dubbo adapter, Among them  init() Function will automatically inject Correlation  filter.Dubbo-Go  Version needs to be ≥1.3.0.Sentinel Dubbo adapter  It will automatically count all  provider  and  consumer  Call to .

• gRPC Application access

import (
 sentinelPlugin "github.com/alibaba/sentinel-golang/adapter/grpc"
 "google.golang.org/grpc"
)

s := grpc.NewServer(grpc.UnaryInterceptor(sentinelPlugin.NewUnaryServerInterceptor()))


• gRPC  Apply to  gRPC  In the initialization code of  Sentinel  Provided  interceptor,Sentinel  in the light of  Server  and  Client  provide  unary  and  streaming  Two kinds of  interceptor, The above code is in  Server  End as an example . The default flow limiting processing logic is to return  Sentinel  Of  BlockError. You can also create  interceptor  Provide customized  fallback  Processing logic .

• Gin Web  Application access

import (
 sentinelPlugin "github.com/alibaba/sentinel-golang/adapter/gin"
 "github.com/gin-gonic/gin"
)

r := gin.New()
r.Use(sentinelPlugin.SentinelMiddleware())


• Gin Web  Apply to  Gin  In the initialization code of  SentinelMiddleware.Sentinel  For each  API route  Make statistics , The resource name is similar to  GET:/foo/:id. The default flow limiting processing logic is to return  429 (Too Many Requests) Error code .

• Micro Application access

import (

 sentinelPlugin "github.com/alibaba/sentinel-golang/adapter/micro"
 "github.com/micro/go-micro/v2"
)

svc := micro.NewService(micro.WrapHandler(sentinelPlugin.NewHandlerWrapper()))


• stay  Go-Micro  In the initialization code of  Sentinel  Provided  wrapper.Sentinel  in the light of  Go-Micro Server  and  Client  provide  wrapper. The above code is in   Server End as an example . The embedded point will extract the service by default  method  As a resource name , The default flow control processing logic is to return  Sentinel  Of  BlockError. You can also create  wrapper  Provide customized  fallback  Processing logic .

How to configure current limiting degradation rules

After connecting through the above methods , You can configure flow control rules 、 Isolation rules 、 Fusing rules 、 Hot rules and other rules ：

• Check the application

After access , You can see the applications we access on the application list page ：

On the app details page , You can also see the overview data of the application ：

You can view the resources registered in the code ：

• Configure flow control rules

Flow control rules can be set for each resource ：

After setting flow control rules , You can see the rejection  QPS  increase 、 adopt  QPS  Reduce ：

Choose flow control protection , You can preset the affordable service  QPS  Traffic , When the traffic reaches the set threshold, immediately intercept the excess requests , Avoid applications being overwhelmed by transient traffic peaks .

• Configure isolation rules

Select isolation protection , You can control the interface or the number of dependent concurrent threads , To ensure the stability of the system . It is generally applicable to scenarios where the internal or downstream dependencies of applications are unstable , For example, slow  SQL、 The response time of downstream applications becomes longer .

• Configure fusing rules

Choose fuse protection , It can monitor the response time or abnormal proportion that the application depends on internally or downstream , When the specified threshold is reached, immediately reduce the priority of downstream dependencies . Within a specified time , The system will not call this unstable resource , Avoid affecting the application , So as to ensure the high availability of the application . When the specified time has passed , Then resume the call to the resource .

• Configure hotspot rules

Select hotspot parameter protection ,AHAS  Statistical parameters will be analyzed , That is, the parameter with a high number of calls in the process of resource call , And limit the flow of resource calls containing hotspot parameters according to the configured hotspot rules , Protect system stability . Hot spots are frequently accessed data . For example, in the following scenarios, it is necessary to count the most frequently accessed  Top  data , And restrict their access .

• For the most frequently purchased goods in a period of time  ID  Limit , Prevent the situation of breaking down the cache and causing a large number of requests to the database .

• For users who visit frequently over a period of time  ID  Limit , Prevent malicious swiping .

MSE  Microservice governance Planning

OpenSergo  Open source project consortium  Sentinel  project , Is making 、 Improve the current limiting degradation standard , It is convenient for different language applications to realize unified microservice governance through unified control surface .

Alibaba cloud micro service engine （MSE） And will gradually support  OpenSergo  standard , Enable microservice developers to use  OpenSergo  To govern microservice applications in different languages .

meanwhile , Alibaba cloud  MSE  It will also explore based on micro service governance , Explore and realize traffic management 、 Current limiting the drop 、 Database governance 、 Message management and other functions , Bring full lifecycle microservice governance to microservice developers .

MSE  First purchase of professional edition of registration configuration center  9  A discount ,MSE  Full specification of cloud native gateway prepaid  9  A discount . Click on

here

, Enjoy the discount ！

Reference link ：

[1]

 

MSE Go SDK：

https://mse-cn-hangzhou.oss-cn-hangzhou.aliyuncs.com/sdk/go/latest/aliyun-mse-go-sdk.zip

[2]

 

12 factor ：

https://12factor.net/zh_cn/