CN106888254B - Kubernetes-based container cloud architecture and interaction method among modules thereof - Google Patents

Abstract

The invention discloses a Kubernetes-based container cloud architecture, which comprises a processor, a mirror image construction module, a data warehouse module, a load balancing module, a service discovery module and a container monitoring module, wherein the mirror image construction module, the data warehouse module, the load balancing module, the service discovery module and the container monitoring module are respectively connected with the processor; the data warehouse module is used for storing and processing data information of the database in the cluster; the load balancing module is used for carrying out load balancing on each computing node in the Kubernetes cluster; the service discovery module is used for acquiring the information of the dynamic change of each computing node in the Kuberntes cluster; and the container monitoring module is used for collecting and displaying the information of the running state of each computing node of the Kubernetes cluster. The invention can be conveniently applied to container systems which need high availability and expansibility.

Description

A Kubernetes-based container cloud architecture and the interaction between its modules method

technical field

The invention relates to the field of cloud computing technology and containers, in particular to a container cloud architecture based on Kubernetes and an interaction method between modules thereof.

Background technique

With the rapid development of the mobile Internet, the number of netizens and the time spent on the Internet have grown rapidly, and the background structure of the website is constantly changing to cope with the ever-increasing demand for access. The design of the server architecture, from the initial deployment of all services such as Web services and database services to a physical server, to the separation of database services and Web services later, so as to improve the performance and security of the server, and then to the use of load balancing technology to The load is distributed to multiple servers to reduce the pressure on a single server. To the nearest automatic scaling method, by monitoring the CPU and memory of the existing server cluster and other indicators, according to the access requirements and the formulated scaling strategy, the system can automatically increase and move. Remove the server node. Compared with the traditional manual addition and deletion of nodes, this method has fast response, low operation and maintenance cost, and high stability.

However, there are many problems in implementing auto-scaling architecture using traditional server and cloud computing technologies, resulting in that auto-scaling methods have not been widely used. First of all, it takes a long time to deploy the business module to the server, and the mobile Internet application requires that the business module must be iteratively launched quickly, and the frequent deployment and update of the business module will waste a lot of manpower and material resources. At the same time, for massive access requests, traditional servers or cloud computing architectures cannot quickly start new service nodes, resulting in access delays or even system crashes. Because the architecture of traditional cloud computing technology basically adopts the virtual machine method, in some sub-fields, its inherent defects lead to unsatisfactory application effects. In recent years, with the popularity of cloud computing technology and the rapid development of related technologies, the implementation of cloud computing technology infrastructure is not limited to a choice of virtual machines.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art, and to provide a container cloud architecture based on Kubernetes, which can be conveniently applied to a container system that requires high availability and scalability.

Another object of the present invention is to provide an interaction method between modules of a Kubernetes-based container cloud architecture.

The object of the present invention is achieved through the following technical solutions:

A Kubernetes-based container cloud architecture includes an image building module, a data warehouse module, a load balancing module, a service discovery module, and a container monitoring module, as well as an image building module, a data warehouse module, a load balancing module, a service discovery module, The server module connected to the container monitoring module, where

Image building module, used to provide image file creation, storage and distribution services;

The data warehouse module is used to store and process the data information of the database in the cluster;

The load balancing module is used to load balance each computing node in the Kubernetes cluster;

The service discovery module is used to obtain information about the dynamic changes of each computing node in the Kubernetes cluster;

The container monitoring module is used to collect and display information about the running status of each computing node in the Kubernetes cluster.

The image building module is a Docker private library module, and uses the Dockerfile packaging technology and the Kubernetes file arrangement template.

The data warehouse module uses hadoop nodes to process Mysql and MongoDB data.

The load balancing module adopts Nginx-Plus proxy tool.

The service discovery module adopts the Etcd storage system.

The server module adopts the CentOS operating system.

Another object of the present invention is achieved through the following technical solutions:

An interaction method between modules of a Kubernetes-based container cloud architecture, including the following steps:

(1) Code update: The developer uploads the new code image to the image building module, and the operation and maintenance personnel downloads the new image from the image building module and loads and runs it;

(2) Load balancing: The requester sends a request, and the load balancing module allocates the received request to any computing node according to the load balancing strategy; the computing node processes the request and returns the result to the load balancing module, and the load balancing module will respond with the result return to the requester;

(3) Service discovery: When a new computing node is added, the information of the newly added computing node will be registered with the service discovery module; when a computing node is removed, the load balancing module is first notified to remove the computing node, and then removed from the service discovery module. In addition to the relevant information of the computing node;

(4) Container monitoring: The monitoring module running in each server module is used as the slave node, and the container monitoring module is used as the master node. The slave node periodically sends the indicator data of the server module where it is located to the master node, and the master node receives the indicators of each slave node. The data is drawn into a graph and displayed on the host browser where the system monitoring module is located;

(5) Data processing: Hadoop is run in the Kubernetes cluster to process massive MongoDB and Mysql data obtained from container applications, so as to achieve the data validity and analyzability of the container system.

Compared with the prior art, the present invention has the following advantages and beneficial effects:

1. LXC container technology has gradually gained a place in the field of cloud computing due to its fast startup speed and good performance. Among them, Docker is an open-source application container engine that allows developers to package applications and dependencies into a portable container and distribute them to Linux machines. Containers use a sandbox mechanism, do not have any interface with each other, and can easily run on hosts and data centers. The Kubernetes used in the present invention, as an engine for managing the lightweight container Docker, can conveniently and quickly create a container and perform full life cycle management of the container, which further promotes the popularization of container technology. The main functions of the Kubernetes container cluster management system include: using Docker to package, instantiate and run applications; run and manage containers across hosts in a cluster; solve communication problems between containers running on different hosts and many more. Going a step further, Kubernetes, as one of the earliest container scheduling solutions based on Docker, is expected to become the most successful container scheduling solution through mature system architecture and powerful horizontal expansion capabilities, and will serve the future cloud computing field.

2. The present invention can be conveniently applied to a container system that requires high availability and scalability, and simultaneously realizes the following functions: (1) Monitor the running status of the CentOS server and its running Docker container; (2) Use Kubernetes technology to dynamically scale according to the monitoring data (3) By building a Docker private Registry warehouse and formulating job management strategies, the continuous integration of Docker images is realized, and the business can be updated without restarting the container.

Description of drawings

FIG. 1 is a schematic structural diagram of a Kubernetes-based container cloud architecture according to the present invention.

FIG. 2 is an architecture diagram of an operation process of an image building module of the container cloud architecture described in FIG. 1 .

FIG. 3 is an operational process architecture diagram of the load balancing module of the container cloud architecture described in FIG. 1 .

FIG. 4 is an architecture diagram of an operation process of the service discovery module of the container cloud architecture described in FIG. 1 .

FIG. 5 is a schematic diagram of the operation process of the container monitoring module of the container cloud architecture described in FIG. 1 .

The labels of the components in the accompanying drawings are as follows: 1-image building module, 2-load balancing module, 3-service discovery module, 4-container monitoring module.

Detailed ways

The present invention will be described in further detail below with reference to the embodiments and the accompanying drawings, but the embodiments of the present invention are not limited thereto.

As shown in Figure 1 to Figure 5, the details are as follows:

(1) Implementation method of image automatic deployment

The image construction module 1 is used to provide an image construction strategy from development to deployment and job definition, mainly supports an automated combination method of image and source code, and is more suitable for timely update of container images during business updates. The process steps of image construction are shown in Figure 2, and the detailed steps are as follows:

(1) Code submission

The developer submits the codes of programming in various languages to the version management warehouse. In the present invention, SVN is used as the version management tool, which is suitable for ensuring the consistency of the code in the project development process of team cooperation.

(2) Define the job

At the same time, the operation and maintenance personnel write the Docker image packaging file Dockerfile suitable for the project and the Kubernetes-based yaml file, and create the corresponding container operation rules in strict accordance with the Dockerfile writing rules and the yaml file arrangement rules.

(3) Build an image file

According to the CI/CD automated deployment process, based on the source code in (1) and the Dockerfile in (2), the image file of the container is created and pushed to the private Docker image repository.

(4) Run the image file

According to the image file in (3) and the yaml file in (2), run the image file in a rationally orchestrated manner in Kubernetes.

The above steps provide image storage and operation services. The image repository uses the private library installation method recommended by Docker, which solves the problem of combining job definitions with source code.

(2) Container system extension method

The load balancing module 2 is used to load balance each Kubernetes node in a dynamically changing cluster; it is implemented by the Nginx-Plus tool, which is a reliable load balancing tool released by Nginx and can be combined in a Kubernetes cluster. , especially suitable for some heavy-load Web systems. Cooperating with Kubernetes, the load balancing capability of Kubernetes can be strengthened, and the scalability of the container system can be improved. The load balancing module is shown in Figure 3.

The service discovery module 3 is used to store the information of each node of the dynamically changing Kubernetes cluster in real time; it uses the Etcd storage system to store the information of each computing node. Etcd is a highly available Key/Value storage system, mainly used for sharing configuration and service discovery. Etcd can quickly and effectively add and remove information about each current Kubernetes node. The service discovery module is shown in Figure 4.

(3) Container monitoring method

The container monitoring module 4 is used to collect, store and display the container running status information of each node of the Kubernetes cluster; CentOS is the most commonly used server operating system in enterprises, and based on this system, the present invention realizes a container monitoring module based on the combination of Zabbix and cAdvisor . Zabbix is an enterprise-level solution that provides distributed system monitoring and network monitoring functions based on a web interface; cAdvisor is a daemon process for collecting, aggregating, processing and outputting container running indicators. performance data. Docker is supported in Zabbix, and container monitoring is implemented using Zabbix Docker Monitoring. The container monitoring system is shown in Figure 5, and the monitoring steps are as follows:

(1) Deploy Zabbix Server on the Kubernetes master node, monitor the master node, import the Web management interface, and distribute to child nodes.

(2) Deploy Zabbix Agent on sub-nodes to realize host monitoring of each sub-node.

(3) Start the cAdvisor container on each Kubernetes node to monitor the performance of the host container.

(4) Web visualization interface display: display the obtained monitoring data on the front end through HTTP and JSON interfaces.

(4) Interaction method of each module

An interaction method between modules of a Kubernetes-based web server architecture, including the following processes:

(1) Code update: A host is used as a Docker private library server. When the business code is updated, the developer automatically uploads the new code image to the host running the private library module. Every time a Kubernetes node is added, the system Pull the new business image from the private library module and load it to run. as shown in picture 2.

(2) Load balancing: The requester sends an http request, and the load balancing module allocates the received request to any Kubernetes node according to the load balancing policy; then the node processes the request and returns the result to the load balancing module, and the load balancing module will respond The result is returned to the requester.

The load balancing module periodically queries the service discovery module (Etcd), and updates and reloads the configuration file of the load balancing module according to the query result. Specifically: the load balancing module periodically pulls the latest Kubernetes node information from the service discovery module, as shown in Figure 3.

(3) Service discovery: It is implemented with Etcd open source tools. When adding a new computing node, it will register the new computing node information with the service discovery module (Etcd); when removing a computing node, first notify the load balancing module to remove the computing node, and then remove the computing node from the service discovery module The relevant information of the node is shown in Figure 4.

(4) Container monitoring: The "active" mode of reporting from the slave node to the master node is adopted. The monitoring module running in each server module is used as a slave node, and one host runs the system monitoring module as the master node. The slave node periodically sends the indicator data of the server module where it is located to the master node, such as CPU ratio and remaining memory. The node receives the indicator data of each slave node and draws it into a graph, which is displayed on the host browser where the system monitoring module is located, as shown in Figure 5.

(5) Data processing: Hadoop is run in the Kubernetes cluster to process a large amount of MongoDB and Mysql data obtained in container applications, so as to achieve the data validity and analyzability of the container system.

(5) Deployment requirements of the method of the present invention

The present invention requires at least four hosts: one host serves as the host of the service discovery module, one host serves as the host of the load balancing module, one host runs the container monitoring module, one host serves as the server that runs the image building module, and the other hosts serve as the host of the load balancing module. Node running the server module. Each host runs its own business logic, and a Kubernetes-based high-availability server cluster is deployed.

The invention discloses a Kubernetes-based container cloud architecture and an interaction method between modules, which can be easily applied to a container system that requires high availability and scalability, and simultaneously realizes the following functions: (1) Monitoring the CentOS server and its operation The running status of Docker containers; (2) Use Kubernetes technology to dynamically scale clusters according to monitoring data; (3) By building Docker private Registry warehouses and formulating job management strategies, the continuous integration of Docker images is realized, and the container does not need to be restarted. Update business. High concurrency and stability of the container system are achieved through the above functions.

The above-mentioned embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the above-mentioned embodiments, and any other changes, modifications, substitutions, combinations, The simplification should be equivalent replacement manners, which are all included in the protection scope of the present invention.

Claims (5)

1. A Kubernetes-based container cloud architecture, characterized in that: the system comprises a mirror image construction module, a data warehouse module, a load balancing module, a service discovery module and a container monitoring module, and also comprises a server module which is respectively connected with the mirror image construction module, the data warehouse module, the load balancing module, the service discovery module and the container monitoring module, wherein the server module is connected with the mirror image construction module, the data warehouse module, the load balancing module, the service discovery module and the container monitoring module

The mirror image construction module is used for providing mirror image file making, storing and distributing services;

the data warehouse module is used for storing and processing data information of the database in the cluster; the data warehouse module processes Mysql and MongoDB data by adopting hadoop nodes;

the load balancing module is used for carrying out load balancing on each computing node in the Kubernetes cluster;

the service discovery module is used for acquiring the information of the dynamic change of each computing node in the Kuberntes cluster; the service discovery module adopts an Etcd storage system;

the container monitoring module is used for collecting and displaying the information of the running state of each computing node of the Kubernetes cluster; the monitoring steps of the container monitoring module are as follows:

(1) deploying a Zabbix Server on a Kubernets main node, monitoring the main node, importing a Web management interface, and distributing the Web management interface to child nodes;

(2) deploying Zabbix agents at the child nodes to realize host monitoring of each child node;

(3) starting a cAdvisor container at each Kubernetes node to realize the monitoring of the performance of the host container;

(4) and displaying the obtained monitoring data on the front end through HTTP and JSON interfaces.

2. The Kubernetes-based container cloud architecture of claim 1, wherein: the mirror image construction module is a Docker private library module and uses Dockerfile packaging technology and Kubernets file arrangement templates.

3. The Kubernetes-based container cloud architecture of claim 1, wherein: the load balancing module adopts an Nginx-Plus agent tool.

4. The Kubernetes-based container cloud architecture of claim 1, wherein: the server module employs a CentOS operating system.

5. The interaction method between the modules of a Kubernetes-based container cloud architecture based on the Kubernetes-based container cloud architecture of any one of claims 1 to 4, comprising the following steps:

(1) updating codes, namely uploading a new code image to an image construction module by a developer, and downloading the new image from the image construction module by an operation and maintenance person and loading for operation;

(2) load balancing, namely, a request end sends a request, and a load balancing module distributes the received request to any computing node according to a load balancing strategy; the computing node processes the request and returns the result to the load balancing module, and the load balancing module returns the response result to the request end;

(3) service discovery, namely registering information of the newly added computing node to a service discovery module when the newly added computing node is added; when a computing node is removed, firstly, a load balancing module is informed to remove the computing node, and then the relevant information of the computing node is removed from a service discovery module;

(4) the container monitoring comprises the steps that a monitoring module running in each server module is used as a slave node, a container monitoring module is used as a master node, the slave node periodically sends index data of the server module to the master node, and the master node receives the index data of each slave node, draws a graph and displays the graph on a host browser where a system monitoring module is located;

(5) and (3) data processing, namely operating hadoop in a Kubernetes cluster, and processing mass MongoDB and Mysql data acquired in the container application.

Images