Limitation du débit d'une API Golang à l'aide de Redis

To put Rate Limiting in simpler words, it is a technique in which we limit the number of requests a user or client can make to an API within a given time frame. You might have encountered in the past getting a "rate limit exceeded" message when you tried to access a weather or a joke API. The are a lot of arguments around why to rate limit an API, but some important ones are to make fair use of it, make it secure, safeguard resources from overload, etc.

In this blog, we will create an HTTP server with Golang using the Gin framework apply a rate limit functionality to an endpoint using Redis and store the total count of the requests made by an IP made to the server in a timeframe. And if it exceeds the limit, we set, we will give an error message.

In case you have no idea what Gin and Redis are. Gin is a web framework written in Golang. It helps to create a simple and fast server without writing a lot of code. Redis it's an in-memory and key-value data store that can be used as a database or for caching capabilities.

Prerequisite

• Familiarity with Golang, Gin and Redis
• A Redis instance (We can use Docker or a remote machine)

Now, let's get started.

To Initialize the project run go mod init for eg, go mod init github.com/Pradumnasaraf/go-redis.

Then let's create a simple HTTP server with Gin Framework then we apply the logic for rate limiting it. You can copy the code below. It's very basic. The server will reply with a message when we hit the /message endpoint.

After you copy the below code, run go mod tidy to automatically install the packages we have imported.

package main

import (
    "github.com/gin-gonic/gin"
)

func main() {
    r := gin.Default()
    r.GET("/message", func(c *gin.Context) {
        c.JSON(200, gin.H{
            "message": "You can make more requests",
        })
    })
    r.Run(":8081") //listen and serve on localhost:8081
} 

We can run the server by executing go run main.go in the terminal and see this message in the terminal.

To test it, we can go to localhost:8081/message we will see this message in the browser.

Now our server is running, let's set up a rate limit functionality for the /message route. We will use the go-redis/redis_rate package. Thanks to the creator of this package, we don't need to write the logic for handling and checking the limit from scratch. It will do all the heavy lifting for us.

Below is the complete code after implementing the rate-limiting functionality. We will understand each bit of it. Just gave the complete code early to avoid any confusion and to understand how different pieces work together.

Once you copy the code run go mod tidy to install all the imported packages. Let's now jump and understand the code (Below the code snippet).

package main

import (
    "context"
    "errors"
    "net/http"

    "github.com/gin-gonic/gin"
    "github.com/go-redis/redis_rate/v10"
    "github.com/redis/go-redis/v9"
)

func main() {
    r := gin.Default()
    r.GET("/message", func(c *gin.Context) {
        err := rateLimiter(c.ClientIP())
        if err != nil {
            c.JSON(http.StatusTooManyRequests, gin.H{
                "message": "you have hit the limit",
            })
            return
        }
        c.JSON(http.StatusOK, gin.H{
            "message": "You can make more requests",
        })
    })
    r.Run(":8081")
}

func rateLimiter(clientIP string) error {
    ctx := context.Background()
    rdb := redis.NewClient(&redis.Options{
        Addr: "localhost:6379",
    })

    limiter := redis_rate.NewLimiter(rdb)
    res, err := limiter.Allow(ctx, clientIP, redis_rate.PerMinute(10))
    if err != nil {
        panic(err)
    }
    if res.Remaining == 0 {
        return errors.New("Rate")
    }

    return nil
}

Let's first directly jump to the rateLimiter() function and understand it. This function asks for an argument that is the IP address of the request which we can obtain via c.ClientIP() in the main function. And we return an error if there is limit is hit otherwise keep it nil. Most of the code is boilerplate code which we took from the official GitHub repo. The key functionality to look closer into here is the limiter.Allow() function. Addr: takes the URL path value for the Redis instance. I am using Docker to run it locally. You can use anything, just make sure you replace the URL accordingly.

res, err := limiter.Allow(ctx, clientIP, redis_rate.PerMinute(10))

It takes three arguments, the first is ctx, the second one is Key, Key (key for a value) for the Redis Database, and the third one is the the limit. So, the function stores the clientIP address as a key and the default limit as the value and reduces it when a request is made. The reason for this structure is that the Redis database needs unique identification and a unique key for storing key-value pairs kind of data, and every IP address is unique in its own way, this is why we are using IP addresses instead of usernames, etc. The 3rd argument redis_rate.PerMinute(10) can be modified as per our need, we can set limit PerSecond, PerHour, etc, and set the value inside parentheses for how many requests can be made per minute/second/hour. In our case, it's 10 per minute. Yes, it's that simple to set.

At last, we are checking if there is a remaining quota of not by res.Remaining. If it's zero we will return an error with the message otherwise we'll return nil. For eg, you can also do res.Limit.Rate to check the limit rate, etc. You can play around and dig deeper into that.

Now coming the main() function:

func main() {
    r := gin.Default()
    r.GET("/message", func(c *gin.Context) {
        err := rateLimiter(c.ClientIP())
        if err != nil {
            c.JSON(http.StatusTooManyRequests, gin.H{
                "message": "you have hit the limit",
            })
            return
        }
        c.JSON(http.StatusOK, gin.H{
            "message": "You can make more requests",
        })
    })
    r.Run(":8081")
}

Everything is almost the same. In the /message route, now every time the route gets hit, we call the rateLimit() function and pass it a ClientIP address and store the return value (error) value in the err variable. If there is an error we will return a 429, that is, http.StatusTooManyRequests, and a message "message": "You have hit the limit". If the person has a remaining limit and the rateLimit() returns no error it will work normally, as it did earlier and serve the request.

That was all the explanation. Let's now test the working. Re-run the server by executing the same command. For the 1st time, we will see the same message we got earlier. Now refresh your browser 10 times (As we set a limit of 10 per minute), and you will see the error message in the browser.

We can also verify this by seeing the logs in the terminal. Gin offers great logging out of the box. After a minute it will restore our limit quota.

本部落格到此結束，我希望您喜歡閱讀，就像我喜歡寫這篇文章一樣。我很高興您能堅持到最後，非常感謝您的支持。我還經常在 X (Twitter) 上談論 Golang 以及其他內容，例如開源和 Docker。你可以在那裡聯絡我。

Ce qui précède est le contenu détaillé de. pour plus d'informations, suivez d'autres articles connexes sur le site Web de PHP en chinois!