应用的一致性问题:在设计之初,就要避免这种问题。不是遇到才去想怎么修数据
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227200936426-2079565353.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227200924433-1566710194.png)
5:架构设计的问题引发
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227201029062-1222990315.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227201704692-1467775797.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227201728084-1888768869.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227201824691-1178610227.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227201912444-390089373.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227202505951-810110085.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227202629039-154860857.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227202938197-1623152836.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227203102918-1383600210.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227203407050-1678407229.png)
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227203548239-932798074.png)
阿里的单元化方案,不会出现跨机房的问题
读比较多,写比较少,微博
![](https://www.icode9.com/i/l/?n=18&i=i-beta/280044/202002/280044-20200227203916011-495647815.png)
业务的妥协:业务的主键不是使用db的自增机制
读多写少,单点写,其它机房部署的db节点,只能是slave
多点接入机房提供服务
RabbitMQ是有ACK的
Kafka设计之初就是高吞吐量,是没有ACK的
kafka 也有ack=all ,消费也可以提交 offset
标签:架构,ACK,db,问题,机房,异地,设计,双活,之初
来源: https://www.cnblogs.com/softidea/p/12374275.html