InfoScale™ 9.0 Replication Administrator's Guide - AIX

How VVR processes application writes

This section helps you understand how application writes are directed when VxVM is not being used, when VxVM is added, and when VVR is added.

When VxVM is not being used, the application writes to a file system placed on a disk partition. In the case of applications or databases on raw devices, the database writes directly to the disk partition, instead of to a file system. In either case, the application, that is, the database or a file system, sends data to the operating system to be written to disk and the operating system communicates directly with the disks.

When VxVM is being used, the applications write to logical devices called volumes, rather than physical disks. A volume is a virtual disk device that appears as a physical disk to applications, such as databases and file systems. However, a volume does not have the limitations of a physical disk.

When VVR is added, it resides between the application and the underlying VxVM volumes. All writes to these replicated volumes are intercepted and replicated to the Secondary host in the order in which they were received at the Primary. Writes are also applied to the local volumes. However, reads are directly processed using the local volumes.

Figure: How application writes are processed when VxVM and VVR are used shows how application writes are processed when VxVM and VVR are used.

Figure: How application writes are processed when VxVM and VVR are used

VVR sends writes to the Secondary in the order in which they were received on the Primary. The Secondary receives writes from the Primary and writes to local volumes.

While replication is active, you should not use the application directly on the data volumes on the Secondary. The application on the Secondary is used only if a disaster occurs on the Primary. If the Primary fails, the application that was running on the Primary can be brought up on the Secondary, and can use the data volumes on the Secondary.

To use the data on the Secondary while the Primary is active, use the snapshot feature to create a version of the data that is not being changed.