Справочник Пользователя для IBM Flex System V7000 Expansion Enclosure 4939H29

Chapter 2. Introduction to IBM Flex System V7000 Storage Node 

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

The contingency capacity is initially set to the real capacity that is assigned when the volume 
is created. If the user modifies the real capacity, the contingency capacity is reset to be the 
difference between the used capacity and real capacity.

A volume that is created with a zero contingency capacity goes offline as soon as it needs to 
expand. A volume with a non-zero contingency capacity stays online until it has been used 
up.

Autoexpand does not cause the real capacity to grow much beyond the virtual capacity. The 
real capacity can be manually expanded to more than the maximum that is required by the 
current virtual capacity, and the contingency capacity is recalculated.

To support the autoexpansion of thin-provisioned volumes, the storage pools from which they 
are allocated have a configurable warning capacity. When the used free capacity of the group 
exceeds the warning capacity, a warning is logged. For example, if a warning of 80% has 
been specified, the warning is logged when 20% of the free capacity remains.

A thin-provisioned volume can be converted to a fully allocated volume using 

 (and vice versa).

IBM Flex System V7000 Storage Node provides a function called 

, which 

enables a volume to have two physical copies. Each volume copy can belong to a different 
storage pool and can be on different managed physical storage systems, which can help 
provide a better level of high-availability to a solution.

When a host system issues a write to a mirrored volume, IBM Flex System V7000 Storage 
Node writes the data to both copies. When a host system issues a read to a mirrored volume, 
IBM Flex System V7000 Storage Node requests it from the primary copy. If one of the 
mirrored volume copies is temporarily unavailable, the IBM Flex System V7000 Storage Node 
automatically uses the alternate copy without any outage for the host system. When the 
mirrored volume copy is repaired, IBM Flex System V7000 Storage Node resynchronizes the 
data.

A mirrored volume can be converted into a non-mirrored volume by deleting one copy or by 
splitting one copy to create a new non-mirrored volume.

The mirrored volume copy can be any type: image, striped, sequential, and thin provisioned 
or not. The two copies can be different volume types.

Using mirrored volumes can also assist with migrating volumes between storage pools that 
have different extent sizes and can provide a mechanism to migrate fully allocated volumes to 
thin-provisioned volumes without any host outages.

You can change the time-out value setting to either 

latency

, which prioritizes low host latency 

(default) or 

redundancy

, which prioritizes redundancy (longer timeout)

Unmirrored volumes: An unmirrored volume can be migrated from one location to 
another by adding a second copy to the wanted destination, waiting for the two copies to 
synchronize, and then removing the original copy. This operation can be stopped at any 
time.