Справочник Пользователя для IBM Flex System V7000 Expansion Enclosure 4939H29

Chapter 9. IBM Flex System V7000 Storage Node Copy Services 

Draft Document for Review January 29, 2013 12:52 pm

This situation provides an always consistent image at the target site and protects your data 
from being inconsistent during re synchronization.

Take a closer look at how Change Volumes might reduce replication traffic. 

Figure 9-64 shows a number of I/Os on the source volume and the same number on the 
target volume, and in the same order. Assuming that this set is the same set of data being 
updated over and over, then these updates are wasted network traffic and the I/O can be 
completed much more efficiently (Figure 9-65). 

In Figure 9-65, the same data is being updated repeatedly, so Change Volumes demonstrate 
significant IO transmission savings, because you only need to send I/O number 16, which 
was the last I/O before the cycling period.

The cycling period can be adjusted by running 

. If a copy does not complete in the cycle period, the next cycle does not start until 

the prior one has completed. It is for this reason that using Change Volumes gives you two 
possibilities for RPO:

��

If your replication completes in the cycling period, then your RPO is twice the 
cycling period.

��

If your replication does not complete within the cycling period, then your RPO is twice the 
completion time. The next cycling period starts immediately after the prior one is finished.

Careful consideration should be put into balancing your business requirements with the 
performance of Global Mirror with Change Volumes. Global Mirror with Change Volumes 
increases the inter-cluster traffic for more frequent cycling periods, so going as short as 
possible is not always the answer. In most cases, the default should meet your requirements 
and perform reasonably well.