一、應用場景

    基(ji)于Nginx的(de)(de)熱(re)點(dian)(dian)(dian)(dian)數(shu)據(ju)調度(du)處(chu)(chu)(chu)理(li)(li)，熱(re)點(dian)(dian)(dian)(dian)節(jie)(jie)點(dian)(dian)(dian)(dian)數(shu)據(ju)負(fu)載均衡處(chu)(chu)(chu)理(li)(li)，減少熱(re)點(dian)(dian)(dian)(dian)節(jie)(jie)點(dian)(dian)(dian)(dian)壓力，提(ti)高(gao)處(chu)(chu)(chu)理(li)(li)和訪(fang)(fang)問效率；每(mei)一個節(jie)(jie)點(dian)(dian)(dian)(dian)的(de)(de)nginx服務(wu)接收(shou)大量的(de)(de)訪(fang)(fang)問，但(dan)是每(mei)個節(jie)(jie)點(dian)(dian)(dian)(dian)處(chu)(chu)(chu)理(li)(li)請(qing)求(qiu)(qiu)都有(you)一個峰值(zhi)，當請(qing)求(qiu)(qiu)數(shu)達到峰值(zhi)時(shi)，后續請(qing)求(qiu)(qiu)的(de)(de)處(chu)(chu)(chu)理(li)(li)效率就會有(you)一定的(de)(de)下降，為了(le)保證請(qing)求(qiu)(qiu)能及時(shi)處(chu)(chu)(chu)理(li)(li)，熱(re)點(dian)(dian)(dian)(dian)節(jie)(jie)點(dian)(dian)(dian)(dian)會觸發(fa)(fa)請(qing)求(qiu)(qiu)調度(du)策略，轉(zhuan)發(fa)(fa)請(qing)求(qiu)(qiu)到非(fei)熱(re)點(dian)(dian)(dian)(dian)節(jie)(jie)點(dian)(dian)(dian)(dian)進行處(chu)(chu)(chu)理(li)(li)，若無非(fei)熱(re)點(dian)(dian)(dian)(dian)節(jie)(jie)點(dian)(dian)(dian)(dian)，則(ze)觸發(fa)(fa)分布式節(jie)(jie)點(dian)(dian)(dian)(dian)策略，備(bei)機節(jie)(jie)點(dian)(dian)(dian)(dian)會啟動Nginx服務(wu)處(chu)(chu)(chu)理(li)(li)，并(bing)接收(shou)熱(re)點(dian)(dian)(dian)(dian)節(jie)(jie)點(dian)(dian)(dian)(dian)轉(zhuan)發(fa)(fa)過來的(de)(de)請(qing)求(qiu)(qiu)，從而提(ti)升訪(fang)(fang)問及處(chu)(chu)(chu)理(li)(li)效率。

二、解決的問題

    解決(jue)大量http請求同一個(ge)Nginx節點繁忙(mang)的(de)問題，做到快速響應。 解決(jue)Nginx集群每個(ge)節點超(chao)負荷(he)接收請求的(de)問題，自動觸(chu)發分布式策略(lve)處(chu)理。 解決(jue)現(xian)有實現(xian)方(fang)案(an)可能存在的(de)單點失效、配置(zhi)不靈活的(de)問題。 

三、部署流程圖

1、客戶端(duan)：用戶通(tong)過客戶端(duan)發起(qi)HTTP請(qing)求。

2、Nginx主機：負責接收客戶端(duan) 的HTTP請求。

3、Nginx備機：等待被Agent喚 醒(xing)的空閑(xian)Nginx節(jie)點。

4、Agent代(dai)理主機(ji)(ji)：負責和(he)Nginx節(jie)點服務進行交互(hu)，包括啟動空閑Nginx備機(ji)(ji)、注銷Nginx節(jie)點服務等。

5、Agent代理備機(ji)(ji)(ji)：作為Agent主機(ji)(ji)(ji)的備用機(ji)(ji)(ji)器，Agent主機(ji)(ji)(ji)宕機(ji)(ji)(ji)的時候，會自動啟動備機(ji)(ji)(ji)作為主機(ji)(ji)(ji)，然后(hou)發(fa)出告警(jing)。

6、Redis緩存：存放(fang)策略(lve)配置(zhi)以及主機IP訪問(wen)量等信息。

三、主要流程圖

主要流程如下：

    Nginx服務接收到(dao)客戶(hu)端HTTP請(qing)(qing)(qing)求(qiu)之后，先讀取redis中該請(qing)(qing)(qing)求(qiu)IP的(de)(de)(de)(de)總訪問量，判斷(duan)如(ru)(ru)果達到(dao)配(pei)置(zhi)(zhi)策略(lve)的(de)(de)(de)(de)峰值，則(ze)(ze)需要獲取峰值策略(lve)判斷(duan)，如(ru)(ru)果配(pei)置(zhi)(zhi)的(de)(de)(de)(de)是控制訪問策略(lve)，則(ze)(ze)拒絕該IP的(de)(de)(de)(de)請(qing)(qing)(qing)求(qiu)；如(ru)(ru)果配(pei)置(zhi)(zhi)的(de)(de)(de)(de)是非(fei)(fei)熱(re)點(dian)(dian)訪問策略(lve)，則(ze)(ze)通知(zhi)Agent轉發(fa)到(dao)非(fei)(fei)熱(re)點(dian)(dian)Nginx節(jie)點(dian)(dian)；如(ru)(ru)果全(quan)部Nginx節(jie)點(dian)(dian)都(dou)達到(dao)峰值，則(ze)(ze)請(qing)(qing)(qing)求(qiu)Agent發(fa)送到(dao)備(bei)用的(de)(de)(de)(de)Nginx節(jie)點(dian)(dian)上。