使用AlwaysON进行DR设计的最佳实践 [英] Best practice for DR design using AlwaysON

问题描述

我有AlwaysOn，有几个AG，每个有3个副本，2个副本在主站点上，第三个在DR站点上。

I have AlwaysOn with a few AGs each has 3 replicas, 2 replicas are on the main site, and the third is on the DR site.

主站点与DR站点之间的距离为1500+英里。

The distance between the main and DR site is 1500+ Miles.

我目前使用同步模式运行所有副本，因为我不想承担在发生灾难时丢失任何数据的风险。在我的情况下，性能不是问题，因为我们的服务器很强大并且吞吐量足够。

I currently run all replicas with synchronous mode as I don't want to take a risk of losing any data in a case of disaster. Performance isn't an issue in my case, as our servers are strong and the throughput is sufficient.

是不是说BP不是同步复制到远处副本？

Is it true to say that it is not the BP to have synchronous replicas to a far replica?

您能否请您参考任何BP文档，该文档将指定同步模式对于远程副本是否正确？

Can you please refer me to any BP documentation which will specify if a synchronous mode is not the right thing for a remote replica?

推荐答案

嗨 
Gal1，

Hi  Gal1,

 

>>是否可以说BP不是具有远程复制副本的同步副本？

>>Is it true to say that it is not the BP to have synchronous replicas to a far replica?

 

基于我的研究，同步提交模式不适用于地理位置较远的可用性副本。

Based on my research, synchronous commit mode is not suitable for geographically distant availability replicas.

 

可用性副本保持同步日志传送。在同步提交模式下，当主副本将日志固化为磁盘时，日志将使用日志扫描程序线程以
的形式发送到可用性副本作为日志块。日志块固化到硬盘，并在可用性副本上重做。在短时间内，可用性副本将solidify和redo进度发送到主副本。当主副本收到 来自辅助副本的
消息（日志已经固化），它完成了事务提交。

Availability replicas are kept in sync by log shipping. In synchronous commit mode, when the primary replica solidifies the logs to disk, logs are sent to the availability replica as log blocks by using the log scanner thread. The log block is solidifies to the hard disk and is redone on the availability replicas. At a short time, the availability replica sends the solidify and redo progress to the primary replica. When the primary replica receives  messages from the secondary replica (the log has been solidified), it completes the transaction commit.

 

  如果辅助副本离主副本太远，则它们之间的网络不稳定，导致主副本无法及时从
辅助副本接收进度消息，这将影响其性能主要副本。这与主服务器和辅助副本之间的网络无关，无论服务器性能如何。

 If the secondary replica is too far away from the primary replica, the network between them is unstable, causing the primary replica to fail to receive progress messages from the secondary copy in a timely manner, which will affect the performance of the primary replica. This is only related to the network between the primary and secondary replicas, regardless of server performance.

 

祝你好运，

Dedmon Dai

Dedmon Dai

这篇关于使用AlwaysON进行DR设计的最佳实践的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！