Shulou(Shulou.com)05/31 Report--

In this issue, the editor will bring you about the main features of HAProxy. The article is rich in content and analyzes and narrates it from a professional point of view. I hope you can get something after reading this article.

In Kubernetes, the Ingress object defines routing rules that specify how to route a client request to a specified service that runs in your cluster. These rules can take into account the unique aspects of the entered HTTP message, including its Host request header and URL path, which will allow you to use data discovery in the request to send traffic from one service to another. That means you can use Ingress objects to define routes for many different applications.

Although Ingress objects define routes, Ingress Controller is the engine that drives them. Ingress Controller is a proxy that sits between the client and the service, and its role is to deliver messages correctly. At present, there are several projects on the market that implement the Ingress Controller specification, each of which has its own advantages. Rancher provides a default controller based on NGINX, but you don't need to confine yourself to it. Rancher has reached a partnership with HAProxy, so you can also choose to use HAProxy Ingress Controller. We tend to think of HAProxy Ingress Controller as a perfect turbocharged engine for Kubernetes, which can fully improve the efficiency of Kubernetes.

Characteristics of HAProxy Ingress Controller

You can find HAProxy in the Rancher App Store, and you can also find the installation details of HAProxy in HAProxy's official documentation (https://www.haproxy.com/documentation/hapee/latest/installation/rancher/). After the setup is complete, HAProxy will automatically listen for and implement Ingress rules. At this point, you can choose to disable NGINX Ingress Controller or keep both Ingress controller running and specify one by name.

The features of HAProxy include the following:

Zero downtime and overload

For many proxy types, including NGINX Ingress Controller, overloading results in a short time window in which the back-end service is not available. In many cases, when HAProxy needs to refresh its configuration, it can avoid reloading altogether.

Its Runtime API allows changes to be made entirely in memory. However, due to the non-disruptive overloading of HAProxy, changes that need to be overloaded will not cause downtime. This means that whenever you add or remove paths, update Secret, or change comments from Ingress rules, traffic will not be affected.

Supercharging performance

Because of HAProxy's focus on performance, you will find that the number of requests that can be processed per second is immediately affected-a significant increase. In addition, through unique algorithms such as elastic binary trees, HAProxy uses fewer resources than other controller.

Observability

Using the Stats page, Runtime API, and raw configuration, you can easily view the configured Pod and associated backends and their health status. The default Ingress controller requires you to install krew to view this information.

HAProxy provides a number of metrics about traffic flowing to your cluster. On the HAProxy Stats page, you will find statistics that track request rates, response times, active connections, success and error responses, and the amount of data passed. This article (https://www.haproxy.com/blog/exploring-the-haproxy-stats-page/) introduces all the metrics provided, which are also exposed through Prometheus endpoints.

HAProxy publishes a detailed log containing request time data, which allows you to pinpoint the slow speed in the request, and the disconnection code will show how and why the request was terminated, as well as the number of active connections in the entire cluster.

Adjustable load balancer

Compared with other Ingress Controller, HAProxy provides more load balancing algorithms, including polling (Round Robin), minimum connection and hash-based algorithms. This choice is important because different types of services perform well in different types of load distribution. For example, a service that maintains a connection for a long time performs better when using the minimum connection algorithm, which checks how busy the server is before sending a new client to the server. You can define it in your Ingress object by adding a comment called haproxy.org/load-balance, using the values listed in the balance document.

Once you enable HTTPS,HAProxy, the end-to-end HTTP/2 is automatically enabled. NGINX supports HTTP/2 on the client side, and HAProxy also supports connecting to your pods through HTTP/2. In addition, HAProxy supports end-to-end streaming for gRPC services.

Enhanced security

Security features such as whitelisting IP addresses and enforcing rate limiting form an important layer of protection. With HAProxy, these functions can be implemented immediately and you can adjust them using annotations. When your cluster hosts multiple services, rate limit is critical, because you certainly don't want one service to take up all the bandwidth.

Overload protection queue

HAProxy's connection queue provides protection against peak traffic. By setting pod-maxconn comments on the Kubernetes service, a set of pod gets the maximum number of concurrent connections, and the number of additional connections is queued to prevent pod overload.

The above describes the six main features of HAProxy, which can help you understand the advantages of HAProxy Ingress Controller more clearly.

These are the main features of HAProxy shared by the editor. If you happen to have similar doubts, you might as well refer to the above analysis to understand. If you want to know more about it, you are welcome to follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.