Thin-aware is the functionality of advertising metro node virtual volumes as thin volumes to hosts. Thin volumes offer more efficiency because the amount of resources used is much smaller than allocated. This benefit of providing only the resource needed exceeds the cost of the virtualization technology that is used. It enables dynamic freeing of storage blocks on storage volumes that have thin support. Thin support enables the mapping of one or more logical blocks to physical blocks, when required. The logical blocks provide the storage address space (logical unit capacity) to hosts. Physical storage is only allocated to the logical unit when it is used. This ensures that the logical unit is allocated less physical storage than it reports as its capacity. The physical blocks can be mapped to the logical blocks when required (on write). Metro node extends multiple thin capabilities that are provided by the arrays that are attached to the back-end.
Thin storage management
Metro node uses some of the management capabilities of the thin-capable arrays in its back-end to detect and address the storage exhaustion issues. When a host stops using the allocated thin storage blocks from the array, the unused blocks are not freed up and they are not returned to the arrays. For example, in a virtual environment where the data in datastores of a virtual machine are stored on a thin volume, and these datastores are deleted or moved, the storage space is not freed up. This behavior can result in an out-of-space issue on the thin volumes. When the thin storage capacity reaches a specific threshold, the storage arrays send out events to the hosts indicating that the storage space is diminishing. In such cases, the hosts can send the SCSI UNMAP command to the metro node virtual volumes to free up the unused space.
NOTE: The UNMAP feature is supported only on the thin-enabled metro node virtual volumes that meet the thin requirements.
Creating thin-enabled virtual volumes lists the thin requirements for a virtual volume.
Thin rebuild
Metro node provides continuous availability and high availability functionalities through its mirroring capability. During the mirroring process, metro node ensures that a thin mirror leg does not turn into a thick leg. Metro node uses its thin rebuild feature to synchronize the data between the mirrors of a RAID-1 device that is built on thin volumes. If the array supports the UNMAP feature, metro node uses the SCSI UNMAP commands to free up space on out of date legs if applicable. If the array does not support the UNMAP feature, metro node writes zeros to blocks that have to be zeroed to preserve thinness. This behavior allows preserving the thinness of the device. Even before the UNMAP support, metro node allowed a metro node administrator to claim a thin storage volume by setting the
thin-rebuild flag. It directs metro node to make efficient use of the space using thin rebuilds.
Metro node supports data mobility features on thin devices. When the migration source or the target is not thin, or the source and the targets are from dissimilar storage-array families, the metro node virtual volume loses its thin properties. In such a case, the virtual volume does not support thin storage management operations. After the migration is completed and committed, the virtual volume will inherit the thin capabilities of the target device.
Migrating thin-capable storage provides you more information on the thin-capable storage migrations.
The following table describes how metro node supports the thin-aware functionalities (based on the understanding of metro node whether the arrays are thin capable).
Table 1. Array thin capability during migration
Functionality
Thin-capable arrays
Arrays that are not thin-capable
Thin provisioning
Discovers the thin volumes on the backend
Automatically sets the
thin-rebuild
flag as part of the storage volume claiming process
Supports the provisioning of thin volumes on the array through VIAS provisioning
Creates the thin-enabled virtual volumes
Supports the manual tagging of the thin volumes with the
thin-rebuild
flag as part of the storage volume claiming process
Thin storage management
Supports the SCSI UNMAP command from the host
Support out-of-space notifications to the host from the last leg that services I/O
Not supported
Thin rebuild
Automatically sets the
thin-rebuild
flag as part of the storage volume claiming process
Uses the SCSI UNMAP command to free up the storage blocks on the out-of-date leg
Supports the manual tagging of the thin volumes with the
thin-rebuild
flag as part of the storage volume claiming process
Uses zero writes as part of the mirror synchronization for the unused blocks
Thin migration
Retains thin storage management capabilities of virtual volume only when migration happens between thin-capable volumes of same storage-array family.
In other scenarios, virtual volume loses thin storage management capabilities during migration, and restores them when migration is committed.
Normal migration behavior with optimization for the unused area.
Data is not available for the Topic
Please provide ratings (1-5 stars).
Please provide ratings (1-5 stars).
Please provide ratings (1-5 stars).
Please select whether the article was helpful or not.
Comments cannot contain these special characters: <>()\