To address this, the logs are placed on fast write cached disks, reducing the disk write time. 为了解决这个问题,把系统放在快速写入高速缓存磁盘上,以减少磁盘写入(DW)时间。
So again, when the disk write completes, it is guaranteed to be both on the production and the backup copy. 所以,当磁盘写入(DW)完成后,可以保证同时在生产和备份副本上完成了写入。
We then ran the workload with varying numbers of users and measured the client response time, probe response time, CPU utilization, disk write times, and disk utilization. 然后,我们使用不同的用户数运行工作负载,并测量客户机响应时间、探测响应时间、CPU利用率、磁盘写时间和磁盘利用率。
It is the disk write speed that causes the bottleneck. 而只有磁盘写入(DW)速率才会造成瓶颈。
Instead, every time a disk write is issued, that write is done on the primary node and sent over to the backup node and written to disk there, as well. 相反,每次有一个磁盘写入(DW)操作发出时,就会在主节点上执行该操作,并且也会被发送到备份节点,在那里的磁盘上完成此写入操作。