Implementing a Global Copy plus Global Mirror topology
This chapter describes the Global Copy plus Global Mirror topology and its use in providing a disaster recovery solution with three sites in three different regions separated by long distances. It includes the following topics:
43.1 Overview of Global Copy plus Global Mirror topology
A topology consisting of a Global Copy and a Global Mirror relationship provides a disaster recovery solution with three sites in three different regions that are separated by long distances. Figure 43-1 shows an example of this topology, where H1:H2 is Global Copy and H1:H3 is Global Mirror. The J3 volumes are the Global Mirror FlashCopy journal volumes.
In the case of a failure at H3, the Global Copy H1:H2 pairs can be put into a Global Mirror session for H1:H2. Therefore, FlashCopy journal volumes at J2 are also required. With support for thin provisioning also for count key data (CKD) environments, the additional capacity needed for the journal volumes is not very large.
As with all types of Multiple Target Peer-to-Peer Remote Copy (PPRC) topologies, there are Multi-Target Incremental Resync relationships (MTIR pairs), also known as Multi-Target Internal Relationships, between the two secondary sites, H2 and H3, as shown in Figure 43-1.
Figure 43-1 Global Copy plus Global Mirror topology
With this topology, a long-distance disaster recovery configuration remains available after an outage at any one of the three sites.
43.2 Creating a Global Copy plus Global Mirror topology
The sections that follow explain the following steps for creating a Global Copy H1:H2 and a Global Mirror H1:H3 topology:
1. Establish PPRC paths between all sites.
2. Establish the H1:H2 Global Copy pairs.
Create FlashCopy H2:J2. (These are required to run Global Mirror H1:H2.)
3. Establish Global Mirror H1:H3:
a. Establish H1:H3 Global Copy pairs.
b. Create FlashCopy H3:J3.
c. Create and start the Global Mirror session for H1:H3.
The details of creating these relationships are similar to creating the Metro Mirror and Global Mirror topology described in 42.2, “Creating a Metro Mirror and Global Mirror topology” on page 497. The one difference is that the H1:H2 relationships are Global Copy rather than Metro Mirror.
The scenarios in this section describe an outage at each of the three sites. Each of these could be either a failure condition or a planned event for testing or maintenance.
To simplify the descriptions, these sites are often referred to as though they were a single volume, but it must be understood that the terms refer to the entire set of volumes being mirrored by PPRC. For example, “Failover H2:H1” means to fail over all of the H2 volumes to their corresponding H1 volumes.
43.2.1 Establish PPRC paths
PPRC paths are required before the PPRC pairs can be established. It is best that paths among all three sites be established so that they are available when they are required.
The procedure for establishing these paths is similar to that for the two Metro Mirror topology that is described in 41.2.2, “Establish PPRC paths” on page 477.
43.2.2 Create H1:H2 Global Copy pairs
After establishing all of the required PPRC paths, create the Global Copy volume pairs H1:H2.
To provide for the possibility of starting Global Mirror H1:H2 in the event of an outage at the Global Mirror remote H3 site, Global Mirror FlashCopy journal volumes J2 are required. These FlashCopy relationships from H2:J2 are created with “inhibit target writes,” “start change recording,” “persistent,” and “no copy” options.
43.2.3 Create H1:H3 Global Mirror
Global Mirror H1:H3 is established in the same manner as described in 42.2, “Creating a Metro Mirror and Global Mirror topology” on page 497.
After both the H1:H2 and H1:H3 relationships have been created, the DS8000 storage system creates the MTIR pairs between H2 and H3, as described in 40.2, “Cascaded PPRC after Multiple Target PPRC failover” on page 467. At this point, there is Global Copy running from H1 to H2, and Global Mirror is running from H1 to H3.
The topology is now as shown in Figure 43-2 on page 518.
Figure 43-2 Global Copy and Global Mirror, including MTIR pairs
43.3 Outage at Global Copy secondary H2
An outage at the Global Copy secondary H2 site can cause the Global Copy H1:H2 pairs to suspend. Because of the design, a suspension of one relationship on a volume does not affect other relationships on the volume, so Global Mirror H1:H3 continues to run and provide a disaster recovery capability, as shown in Figure 43-3.
Figure 43-3 Outage at Global Copy secondary H2
When site H2 is recovered, the Global Copy H1:H2 pairs are resumed and data is copied from H1 to H2 again.
43.4 Outage at Global Mirror H3
An outage at the Global Mirror remote site H3 causes the loss of the Global Mirror disaster recovery capability.
The Global Copy H1:H2 pairs can be placed into a Global Mirror session to restore the disaster recovery capability.
43.4.1 Start Global Mirror H1:H2
Global Copy for H1:H2 remains active. The disaster recovery capability can be restored by converting the Global Copy H1:H2 to Global Mirror. This conversion requires removing the H1:H3 Global Copy pairs from the H1:H3 Global Mirror session, removing the GM H1:H3 session, and adding the Global Copy pairs into a new H1:H2 Global Mirror session.
Because the H1:H2 Global Copy pairs have already been running and transferring data from H1 to H2, there is no additional data transfer required. The H1:H2 pairs are already past the first round of copy, so they will be immediately joined to the session, and consistency groups will begin to be formed at the H2 site.
The resulting configuration is shown in Figure 43-4.
Figure 43-4 Global Mirror H1:H2 after outage at H3
43.4.2 H3 recovered
When the H3 site is recovered, the Global Copy H1:H3 pairs are resumed, resulting in the configuration shown in Figure 43-5 on page 520.
This is the same topology as in the initial starting case with Global Copy plus Global Mirror. However, Global Mirror is now running from H1 to H2 rather than from H1 to H3. If desired, the Global Mirror can be moved back to H1:H3 with a steps similar to those described previously.
Figure 43-5 Resume Global Copy H1:H3
43.5 Outage at H1
In the case of an outage at the H1 site, a recovery at the remote Global Mirror H3 site is required. This requires a Global Mirror recovery and restart of production at the disaster recovery site, H3.
With Multiple Target PPRC, the MTIR pairs between H2 and H3 are used to perform an Incremental Resynchronization between the two sites.
The initial configuration is shown in Figure 43-6, where a failure occurred at the H1 site.
Figure 43-6 Global Copy plus Global Mirror topology
43.5.1 Recover at Global Mirror remote H3 site
The Global Mirror recovery at the H3 site is the same as in a non-Multiple Target PPRC environment. Further information about the process for this is described in Part 5, “Global Mirror” on page 227.
43.5.2 Restore Global Mirror disaster recovery capability
After recovering at the remote Global Mirror location, the disaster recovery protection of Global Mirror can be restored.
Failover H2:H1
A PPRC failover H2:H1 causes H2 to be converted from a secondary of H1 to become a suspended primary to H1. This is required so that H2 can become a secondary of H3.
Incremental Resynchronization for H3:H2
The Multiple Target PPRC configuration created MTIR pairs between H2 and H3. As with the previously described Multiple Target PPRC configurations, the MTIR pairs can be changed to active pairs with a PPRC failback command. In this case, the recovery is performed at the H3 site, and Global Copy is restored with a failback H3:H2 command that specifies Global Copy.
43.5.3 Start Global Mirror H3:H2
A Global Mirror session is started for H3:H2, and then the H3:H2 Global Copy pairs are added to this Global Mirror session. This results in the configuration are shown in Figure 43-7.
Figure 43-7 Restore Global Mirror capability
At this point, the disaster recovery protection of Global Mirror as been restored, with H2 acting as the remote recovery site.
43.5.4 H1 recovered
When the H1 site has been recovered, Global Copy H3:H1 can be resumed with a PPRC failback H3:H1. This results in the configuration are shown in Figure 43-8.
Figure 43-8 Failback H3:H1, Global Copy
The original topology of Global Copy plus Global Mirror has now been restored. The host I/O is running at the H3 site rather than the original H1 site. If you want to return to the H1 site, you can follow a similar sequence of steps to do a planned move back to the H1 site location.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.