Service discovery
In k8s, we usually will more than make an application instance, and also the corresponding multiple pod, if through the pod IP access to the pod service will be hard to manage. Kubernetes provides the concept of service can be accessed through the VIP pod services, but in the use of the time there is a problem: how do you know an application of the VIP?We have two application, for example, an APP, one is the DB, manage each application using the rc, and provide services through service exposed the port.APP needs to connect to the DB application, we only know the name of a DB application, but do not know the VIP address.
 
The simplest way from kubernetes provides query API.But this is a bad practice, every application must first write queries at start-up time rely on the service logic, this itself is repeat and increases the complexity of the application;Second leading to applications rely on kubernetes, will not be able to deploy and run separately (of course, if by adding configuration option is also can be done, but it is increase in degrees).
 
At first, the method of kubernetes adopted docker used - environment variables.Start each pod, will be put through the environment variable is set all service of IP and port information, so that the application of the pod can be read by the environment variable to rely on the service address information.Matching relation between service and environment variables this way has a certain specification, use rise also relatively simple, but there is a big problem: rely on the service must be started in the pod existed before, otherwise will not appear in the environment variable.
 
A more ideal solution is: application can directly use the name of the service, don't need to care about it the IP address of the actual, in the middle of the conversion can be done automatically.DNS is name and IP conversion between the function of the system, so kubernetes is also provides the DNS method to solve this problem.
 
DNS Service
The DNS service is not independent of the system, but an addon, as a plug-in to install, not kubernetes cluster must (but very recommended installation).Can take it as a run on the application of the cluster, it's just the application is special.
There are two kinds of DNS configuration mode, using etcd + kube2sky + skydns way before version 1.3, can be used after 1.3 kubedns + dnsmasq way.
 

kube2sky mode

    This model mainly has three containers in operation:
  • Kube2sky is responsible for the continuous monitoring k8s apiserver, once has the service creation, will obtain the service IP, and stored in the etcd
  • Etcd in the form of the key - value is used to store the service name and the corresponding "ClusterIP"
  • SkyDNS: according to the data of the etcd, external DNS query service

Service discovery
In k8s, we usually will more than make an application instance, and also the corresponding multiple pod, if through the pod IP access to the pod service will be hard to manage. Kubernetes provides the concept of service can be accessed through the VIP pod services, but in the use of the time there is a problem: how do you know an application of the VIP?We have two application, for example, an APP, one is the DB, manage each application using the rc, and provide services through service exposed the port.APP needs to connect to the DB application, we only know the name of a DB application, but do not know the VIP address.
 
The simplest way from kubernetes provides query API.But this is a bad practice, every application must first write queries at start-up time rely on the service logic, this itself is repeat and increases the complexity of the application;Second leading to applications rely on kubernetes, will not be able to deploy and run separately (of course, if by adding configuration option is also can be done, but it is increase in degrees).
 
At first, the method of kubernetes adopted docker used - environment variables.Start each pod, will be put through the environment variable is set all service of IP and port information, so that the application of the pod can be read by the environment variable to rely on the service address information.Matching relation between service and environment variables this way has a certain specification, use rise also relatively simple, but there is a big problem: rely on the service must be started in the pod existed before, otherwise will not appear in the environment variable.
 
A more ideal solution is: application can directly use the name of the service, don't need to care about it the IP address of the actual, in the middle of the conversion can be done automatically.DNS is name and IP conversion between the function of the system, so kubernetes is also provides the DNS method to solve this problem.
 
DNS Service
The DNS service is not independent of the system, but an addon, as a plug-in to install, not kubernetes cluster must (but very recommended installation).Can take it as a run on the application of the cluster, it's just the application is special.
There are two kinds of DNS configuration mode, using etcd + kube2sky + skydns way before version 1.3, can be used after 1.3 kubedns + dnsmasq way.
 

kube2sky mode

    This model mainly has three containers in operation:
  • Kube2sky is responsible for the continuous monitoring k8s apiserver, once has the service creation, will obtain the service IP, and stored in the etcd
  • Etcd in the form of the key - value is used to store the service name and the corresponding "ClusterIP"
  • SkyDNS: according to the data of the etcd, external DNS query service
        
                example:
                        There is a DB, a APP server, APP server needs to connect DB for data reading and writing
                       1.DB server, through the RC created a pod, and at the same time to create a service for integration of the pod,DB service named DB_server.This is accomplished by k8s - apiserver,Will be a cluster IP for DB service distribution,for example: 192.168.20.3,ClusterIP can only be used for use within the cluster.
                       2.kube2sky listening to APIserver operations, access to the service name: DB_server and clusterIP: 192.168.20.3, and will write etcd.
                       3.APP server configured in the DB address for 'DB_server', this is the name of DB service,Use the service name instead of IP
                       4.Pod will send service name: DB_server to skyDNS, loading from etcd skyDNS service name corresponding IP, returned to the Pod
 
kubeDNS mode
    This mode, kubeDNS container replace the function of the original three container, it will be to monitor apiserver and put all the service and the result of the endpoints using appropriate data structure stored in memory, and external DNS query service.

    • KubeDNS: to provide the original kube2sky + etcd + skyDNS function, can provide DNS query service
    • A lightweight DNS service software, can provide DNS cache function.KubeDNS mode, dnsmasq in memory set aside a block size (the default is 1g), save the current most commonly used DNS query record, if there is no to find records in the cache, it will into kubeDNS query, and the results are cached

                    
                 example:
                 1.DB server, through the RC created a pod, and at the same time to create a service for integration of the pod,DB service named DB_server.This is accomplished by k8s - apiserver,Will be a cluster IP for DB service distribution,for example: 192.168.20.3,ClusterIP can only be used for use within the cluster.
                  2. kubeDNS  listening to APIserver operations,access to the service name: DB_server and clusterIP: 192.168.20.3, and use tree structure write cache.
                  3. dnsmasq:DNS rules obtained through kubedns container, in a cluster to provide DNS query service, equivalent to the DNS server.
                  4. APP server configured in the DB address for 'DB_server', this is the name of DB service,Use the service name instead of IP
                  5. Pod will send service name: DB_server to dnsmasq,dnsmasq query to the service of the corresponding IP is returned to the pod, if not checked, will contact the kubeDNS,If kubeDNS no record, it will return an error report to the APP server pod.
                example:
                        There is a DB, a APP server, APP server needs to connect DB for data reading and writing
                       1.DB server, through the RC created a pod, and at the same time to create a service for integration of the pod,DB service named DB_server.This is accomplished by k8s - apiserver,Will be a cluster IP for DB service distribution,for example: 192.168.20.3,ClusterIP can only be used for use within the cluster.
                       2.kube2sky listening to APIserver operations, access to the service name: DB_server and clusterIP: 192.168.20.3, and will write etcd.
                       3.APP server configured in the DB address for 'DB_server', this is the name of DB service,Use the service name instead of IP
                       4.Pod will send service name: DB_server to skyDNS, loading from etcd skyDNS service name corresponding IP, returned to the Pod
 
kubeDNS mode
    This mode, kubeDNS container replace the function of the original three container, it will be to monitor apiserver and put all the service and the result of the endpoints using appropriate data structure stored in memory, and external DNS query service.

    • KubeDNS: to provide the original kube2sky + etcd + skyDNS function, can provide DNS query service
    • A lightweight DNS service software, can provide DNS cache function.KubeDNS mode, dnsmasq in memory set aside a block size (the default is 1g), save the current most commonly used DNS query record, if there is no to find records in the cache, it will into kubeDNS query, and the results are cached

                    
                 example:
                 1.DB server, through the RC created a pod, and at the same time to create a service for integration of the pod,DB service named DB_server.This is accomplished by k8s - apiserver,Will be a cluster IP for DB service distribution,for example: 192.168.20.3,ClusterIP can only be used for use within the cluster.
                  2. kubeDNS  listening to APIserver operations,access to the service name: DB_server and clusterIP: 192.168.20.3, and use tree structure write cache.
                  3. dnsmasq:DNS rules obtained through kubedns container, in a cluster to provide DNS query service, equivalent to the DNS server.
                  4. APP server configured in the DB address for 'DB_server', this is the name of DB service,Use the service name instead of IP
                  5. Pod will send service name: DB_server to dnsmasq,dnsmasq query to the service of the corresponding IP is returned to the pod, if not checked, will contact the kubeDNS,If kubeDNS no record, it will return an error report to the APP server pod.

kubeDNS workflow(service registration and discovery)的更多相关文章

  1. Open-Source Service Discovery

    Service discovery is a key component of most distributed systems and service oriented architectures. ...

  2. Spring Boot Dubbo Dubbok spring cloud

    比较spring cloud和dubbo,各自的优缺点是什么 - 趁年轻再疯狂一次吧 - CSDN博客 https://blog.csdn.net/u010664947/article/details ...

  3. springcloud情操陶冶-初识springcloud

    许久之前便听到了springcloud如雷贯耳的大名,但是不曾谋面,其主要应用于微服务的相关架构.笔者对微服务并不是很了解,但其既然比较出众,遂也稍微接触研究下 springcloud特性 sprin ...

  4. 微服务领域是不是要变天了?Spring Cloud Alibaba正式入驻Spring Cloud官方孵化器!

    引言 微服务这个词的热度自它出现以后,就一直是高烧不退,而微服务之所以这么火,其实和近几年互联网的创业氛围是分不开的. 与传统行业不同,互联网企业有一个特点,那就是市场扩张速度非常之快,可能也就是几天 ...

  5. springcloud

    基本术语 1.服务器 服务器:是提供计算服务的设备.由于服务器需要响应服务请求,并进行处理,因此一般来说服务器应具备承担服务并且保障服务的能力.服务器的构成:包括处理器.硬盘.内存.系统总线等,和通用 ...

  6. 微服务之Spring cloud

    微服务 Spring cloud Spring Cloud provides tools for developers to quickly build some of the common patt ...

  7. Spring Cloud简介

    一.本文介绍 Web应用由最早的单体应用发展成为集群式的部署,再到现在的分布式系统.尤其是这两年分布式相关的技术发展的很快,一方面是以Dubbo为代表的,另一方面则是以Spring Cloud系列为代 ...

  8. Spring Cloud Summary

    Spring Cloud Summary https://cloud.spring.io/spring-cloud-static/Finchley.RC1/single/spring-cloud.ht ...

  9. 搭建Spring Initializr服务器

    前言 按照网上很多教程,出错特别多.首先是GitHub和maven仓库的网络环境比较差,踩了很多坑:其次是SpringInitializr更新迭代几个版本,0.7.0我也没能弄成功.索性就用了旧版本0 ...

随机推荐

  1. LC 413. Arithmetic Slices

    A sequence of number is called arithmetic if it consists of at least three elements and if the diffe ...

  2. flask_sqlalchemy的session线程安全源码解读

    flask_sqlalchemy是如何在多线程中对数据库操作不相互影响 数据库操作隔离 结论:使用scoped_session实现数据库操作隔离 flask的api.route()接收一个请求,就会创 ...

  3. spring boot configuration annotation processor not found in classpath

    <dependency> <groupId> org.springframework.boot </groupId> <artifactId> spri ...

  4. 基于OpenCV的三维数据点的曲面重构_MySurefaceReconstruction

    在Opencv中有个Viz模块,可以显示三维物体,还可以实现三维动画,本来是很好的东东,但是里面的函数.类的说明太过简单,始终不得要领.不过其中一个扩展功能非常好,就是你可以在vtk中设计自己的模型类 ...

  5. Day05:集合操作——线性表(二) / 查找表 / 文件操作——File(一)

    文件操作:https://www.cnblogs.com/wsnb8/p/11403626.html

  6. SQL优化手段

    一.建立索引 要尽量避免全表扫描,首先应考虑在 where 及 order by 涉及的列上建立索引(order by desc会非常影响效率). 二.避免在建立索引的字段进行计算操作. 三.使用预编 ...

  7. 山东省第十届ACM省赛参赛后的学期总结

    5.11,5.12两天的济南之旅结束了,我也参加了人生中第一次正式的acm比赛,虽然是以友情队的身份,但是我依旧十分兴奋. 其实一直想写博客来增加自己的能力的,但是一直拖到现在,正赶上老师要求写一份总 ...

  8. 【AMAD】coconut -- 简单,优雅,pythonic的函数式编程语言

    动机 简介 个人评分 动机 作者的话: 我喜欢函数式编程,我认为函数式编程提供了一个更自然的方式来思考问题,并且代码也更加优雅,易读.不过如果你看过前20个最受欢迎的编程语言,你会发现没有一个式函数式 ...

  9. Leetcode之动态规划(DP)专题-64. 最小路径和(Minimum Path Sum)

    Leetcode之动态规划(DP)专题-64. 最小路径和(Minimum Path Sum) 给定一个包含非负整数的 m x n 网格,请找出一条从左上角到右下角的路径,使得路径上的数字总和为最小. ...

  10. redhat 7 防火墙配置

    没有iptables 用systemctl stop firewalld