< Previous  | TOC | Index | Next  >

Example 1—Decision support using the traditional snapshot feature and the vxibc utility

1. On the Secondary, register the application name dss_app and get ready to receive the IBC message. The command to break-off the snapshot plex when the IBC is received is specified with the vxibc regrecv command, as follows:

   # vxibc -g hrdg regrecv dss_app hr_rvg vxrvg snapshot hr_rvg

2. On the Primary, put the application that is using the Primary data volumes hr_dv01 and hr_dv02 into a consistent state.

   ---

     Note   Putting the application in a consistent state involves flushing all the buffers and pending transactions. For example, the file system can be brought into a consistent state by synchronizing the volumes using the VxFS specific sync command. In Oracle, the application can be brought into a consistent state by putting it in a hot-backup mode.

   ---

3. On the Primary, register the application name dss_app and send the IBC message using the following command:

   # vxibc -g hrdg regsend dss_app hr_rvg

   When the Secondary receives the IBC message, replication is frozen and no more data is written to the secondary data volumes. The vxrvg snapshot command specified in step 1 then breaks-off the snapshot plexes from the Secondary data volumes, and replication starts again. When the vxibc commands complete on both the hosts, the application name is unregistered.

4. On the Primary host, resume the application if it was suspended in step 2.

   If the RLINK is asynchronous and behind there may be a delay between the vxibc regsend and vxrvg snapshot command. During this delay, the application is running.

5. On the Secondary, use the snapshot data volumes SNAP-hr_dv01 and SNAP-hr_dv02 for running the DSS application, that is, for off-host processing.
6. When the application completes, reattach the snapshot plexes to the data volumes using the following command:

   # vxrvg -g hrdg snapback hr_rvg

   The reattach destroys the SNAP volumes and reattaches the snapshot plexes to their original volumes. If you have enabled FR on these volumes, only the blocks that were changed by the off-host processing application are resynchronized.