HP P6000 Replication Solutions Manager User Guide (T3680-96089, October 2012)

When logging, the size automatically increases in proportion to the host data written to the DR
group, up to a preset maximum. By default, the maximum size is determined by the controller
software. However, you can manually set the maximum.
Log size example
Maximum size
[varies]
^
^
^
Initial (minimum) size
• The actual size and the requested size of a log are reported. When these differ, it indicates
the requested size change cannot be made until the log state changes.
• In synchronous write mode, the log is automatically returned to its minimum size when a merge
is completed or a full copy is performed. In enhanced asynchronous write mode, the log is
not returned to its minimum size when a merge is completed or when a full copy is performed.
• Sometimes a log can become much larger than is typically needed. In this case, HP recommends
that you edit the DR group properties and make the log smaller.
Default maximum size
The default maximum size depends on the controller software version and equals the DR group
size multiplied by a factor. For example, if a DR group contains virtual disks with a total of 10 GB
and the factor is 1x then the default maximum size is 10 GB. For specific factors, see Controller
software features - remote replication.
Log size and disk group occupancy warnings
The size of a log disk can cause a disk group occupancy warning. See disk group occupancy
alarm level. If this happens, consider the following options:
• Temporarily stop host I/O to the virtual disks in the impacted DR group and plan necessary
changes.
• Allow the log to completely fill and trigger an automatic full copy, or manually force a full
copy. See Full copy mode. This returns the log to its minimum size.
• Add storage to the disk group that the log disk is in.
• Delete unneeded virtual disks or containers in the disk group to create more free space.
Logging
The process of storing host write-transactions to the virtual disks in a source DR group, in the exact
order received, is called logging. The details of the logging process depend on remote replication
write mode. See Write mode.
Logging when in enhanced asynchronous write mode
When using enhanced asynchronous write mode, logging occurs continuously, even if replication
is not suspended or the destination is available. In effect, the log continuously acts as a FIFO buffer,
accepting host write-transactions in and replicating them out to the destination.
In many cases, an asynchronous log can buffer up to several days of host write-transactions,
depending on the log size and the transaction rate.
DR group concepts 87