Chapter 22. Global Mirror overview 247
In synchronous data replication methods such as Metro Mirror, an application write always
goes through the following four steps; see Figure 22-1 on page 246:
1. Write the data to the primary storage disk subsystem cache and present channel end to
free up the channel for further I/O. Note this does not end the I/O and does not present an
I/O complete to the application.
2. Replicate the data from the primary storage disk subsystem cache to the secondary
storage disk subsystem cache.
3. Acknowledge to the primary storage disk subsystem that data successfully arrived at the
secondary storage disk subsystem.
4. Present device end to acknowledge successful I/O completion to the server, which is
presented to the application and concludes this I/O.
Now the next I/O that depends on the successful completion of the previous I/O can be
issued.
When you have dependent writes across multiple storage disk subsystems, synchronous
data replication alone does not guarantee that you can restart applications at the secondary
site without doing some previous data recovery.
Consider a database environment that spreads across multiple storage disk subsystems at
the local (primary) site. Further assume that the remote copy volume pairs are defined with
CRIT(HEAVY), which used to be CRIT(YES), and the remote copy paths are not defined with
the CGROUP parameter. Figure 22-2 illustrates what happens in this scenario during a
disaster, and in the absence of any automation software in place to handle I/O exceptions.
Figure 22-2 Synchronous data replication and unnecessary recovery after primary site failure
Replicate
Primary
Primary
A
Primary
Primary
A
2D00
Primary
Primary
A
Primary
B3
3E00
Subsystem
Database
Primary
A
2C00
Primary
Primary
A
Primary
A3
2E00
Primary
Primary
A
Primary
A1
2C00
B1
3C00
Synchronous
Replicate
Synchronous
Storage Disk
Subsystem 5
Storage Disk
Subsystem 6
Storage Disk
Subsystem 1
Storage Disk
Subsystem 2
1
2
3
Replicate
Primary
Primary
A
Primary
Primary
A
2D00
Primary
Primary
A
Primary
B2
3D00
Primary
Primary
A
Primary
A2
2D00
Synchronous
Storage Disk
Subsystem 3
Storage Disk
Subsystem 4
3
1
Secondary
Primary
RECON'
Log'
DB'
RECON
Log
DB
Subsystem
Database
Recover A2
6
5
Recover A2
4
Switch over to secondary site
Restart DB Subsystem