Unplanned outage of local (H1) with production move to remote (H3 GM secondary)

Edit online

In this scenario, there is an unplanned outage for Site 1 (H1). Site 2, however, does not have the resources to run production, so production will be moved to Site 3 during the outage at Site 1. When the outage is complete, production can then be moved back to Site 1 and resume the full three-site configuration.

About this task

Follow these steps to carry out this scenario.

Procedure for Multi-Target Metro Mirror - Global Mirror session type

Edit online

Procedure

  1. Start H1->H2 H1->H3

    Issue this command to begin multi-target replication when the production I/O is running on H1. When the session is in a Prepared state, both sites 2 and 3 have a consistent copy of the data.

  2. Unplanned outage of H1 and H2
    This event causes an unplanned outage of H1 and H2 which causes the session to go suspended.
    Note: If the session did not automatically go into a Suspended state, issue the Suspend command. The command fail if it is unable to talk to the H1 volumes, but the session will be forced into the Suspended state, which will allow the Recover command.
  3. Recover H3

    This command makes H3 Target Available. The session will be severe/Target Available. When the session state is Target Available at H3, production I/O can be moved to H3.

    Note: Once Site 1 is brought back up, perform the following step to begin the process of returning production to Site 1.
  4. StartGC H3->H1->H2, StartGC H3->H2->H1, StartGM H3->H1->H2or StartGM H3->H1->H2
    Issue this command when H1 and H2 are available and ready to be brought back into the configuration. These commands start replication back from the H3 site to the H1 and H2 sites.
    Note: The StartGM command is only available on session types allowing GM back to the H1 or H2 sites. This allows for disaster recovery capabilities while running at site 3. If the session type only supports the StartGC command at Site 3, then disaster recovery will not be available until moving production back to Site 1 or Site 2.
    Note: When you are ready to switch back to Site 1, stop production I/O on H3 before continuing to the next step. When running in Global Copy mode, it is recommended that you do the next step once all pairs are close to being synchronized, to minimize the time it will take to suspend the session consistently.
  5. Suspend

    This command suspends the relationships to Site 1 and Site 2 consistently. When this step is complete, the session state should be suspended and recoverable.

  6. Recover H1

    This command makes H1 Target Available. When the session state is Target Available at H1, run production I/O to H1.

  7. Start H1->H2 H1->H3

    This command restores the original three-site configuration. Once the session reaches a Prepared state, a consistent copy of the data will reside on Site 2 and Site 3.

Procedure for Cascaded Metro Global Mirror session type

Edit online

Procedure

  1. Start H1->H2->H3

    Issue this command to begin Metro Global Mirror on your system when the production I/O is running on H1. When the session is in a Prepared state, both sites 2 and 3 have a consistent copy of the data.

  2. Unplanned outage of H1 and H2

    This event causes an unplanned outage of H1 and H2 which causes the session to go suspended. It is recoverable if the session was in a Prepared state when the outage occurred. If the state was Preparing, the session might be recoverable at Site 2, but will not be recoverable at Site 3.

  3. Suspend

    You might need to issue a Suspend at this point. Although the suspend might fail, it will put the session in a suspended state, and therefore allow the Recover H3.

  4. Recover H3

    This command makes H3 Target Available. The session will be severe/Target Available. When the session state is Target Available at H3, production I/O can be moved to H3.

    Note: Once Site 1 is brought back up, perform the following step to begin the process of returning production to Site 1.
  5. Start H3->H1->H2

    Issue this command when H1 and H2 are available and ready to be brought back into the configuration. Changes from production at the remote site (H3) flow from the local site (H1) to the intermediate site (H2). The H3-H1 and H1-H2 pairs are Global Copy. The session will in the preparing state.

    Note: While the session is Preparing, you can decrease the amount of time that production I/O has to be stopped by waiting for the progress on the H3-H1 role pair to be at or near 100%. The next step will drive H3-H1 to sync, which depending on the percentage of the copy that is complete, can take a significant amount of time. When you are ready to switch back to Site 1, stop production I/O on H3 before continuing to next step.
  6. Suspend

    This command suspends the Global Copy relationship between H3>H1 and H1>H2. When this step is complete, the session state should be suspended and recoverable.

  7. Recover H1

    This command makes H1 Target Available. When the session state is Target Available at H1, run production I/O to H1.

  8. Start H1->H2->H3

    This command restores the original three-site configuration. Once the session reaches a Prepared state, a consistent copy of the data will reside on Site 2 and Site 3.