Hardware accelerated compression and controller-based compression
Capacity saving and accelerated compression functions
The VSP G series and VSP F series storage systems provide the following functions to make efficient use of user capacity:
- Capacity saving The capacity saving function includes data deduplication and data compression. Capacity saving enables you to reduce your bitcost for the stored data by deduplicating and compressing the data. Data deduplication and compression are performed by the controllers of the storage system.
- Accelerated compression The accelerated compression function enables you to reduce your bitcost for the stored data by allowing you to take advantage of the compression function in the FMD drives. Accelerated compression allows you to assign FMD capacity to a pool that is larger than the physical capacity of the FMD parity groups. The data access performance of the storage system is maintained when the accelerated compression function is used, as the compression engine is offloaded to the FMD drives.
The following table lists the combinations of applying accelerated compression and capacity saving and describes the functionality of each combination.
Combination |
Functionality |
Use only accelerated compression |
Data compression is performed by the FMD drives. The storage controller does not perform the compression. I/O performance is not affected because there is no overhead due to compression processing by the storage controller. |
Use only controller-based compression |
The storage controller compresses data and stores the compressed data in the pool. Use controller-based compression for drives other than FMD and for encrypted drives. Using accelerated compression and controller-based compression at the same time results in compression being performed in the FMD drives only. Because of differences in storing and managing data by the storage controller, performance is lower than when only accelerated compression is used. Therefore, when accelerated compression can be enabled, you should use accelerated compression instead of controller-based compression. Use controller-based compression only when accelerated compression cannot be enabled. |
Use accelerated compression and deduplication and compression See Example 1 below |
When multiple copies of identical data are stored in a pool, the storage controller keeps only one copy (deduplication). For compression, the storage controller detects that accelerated compression is enabled* and uses it instead of controller-based compression. *Accelerated compression must be enabled for all parity groups in the pool. |
Use deduplication and compression See Example 2 below |
The storage controller performs compression and deduplication processing. The storage controller has the largest overhead of the capacity saving processing. |
- When BED encryption is not being utilized, use accelerated compression.
- When accelerated compression is being utilized, do not enable controller-based compression.
Example 1: Applying deduplication and accelerated compression
Example 2: Applying deduplication and compression (capacity saving function)
- Saving Effect: If the capacity saving function is used, the saving ratio is calculated to include user data, meta data and garbage data (that is generated by the system). If the data volume used capacity before the saving is smaller than the pool used capacity, the saving ratio might be lower and the saving capacity might be a hyphen (-) as an invalid value. If data in DP-VOLs is deleted by the types of tasks listed below, the saving ratio might be lower than expected because the non-formatted data exists in FMD:
- Deleting DP-VOLs
- Formatting DP-VOLs
- Initializing duplicated data in a pool
- Software Saving: If the capacity saving function is used, the saving ratio is calculated with user data, meta data and garbage data (that are generated by the system). If the data volume used capacity before the saving is smaller than the pool used capacity, the saving ratio might be lower and the saving capacity might be a hyphen (-) as an invalid value. If you initialize the duplicated data in a pool, the saving effect might be different than expected, at least until after the LDEV format processing has completed. Verify the saving effect only after the duplicated data in a pool has finished initializing.
- FMD Saving: If data in DP-VOLs is deleted by the types operations listed below, the saving ratio might be different than expected because the non-formatted data exists in FMD:
- Deleting DP-VOLs
- Formatting DP-VOLs
- Initializing duplicated data in a pool
For details about the formatted pool capacity or progress, see the item for Free + Reserved in the GUID-B12C15FC-56F9-4EA4-A976-D1A4691DAAD5#GUID-B12C15FC-56F9-4EA4-A976-D1A4691DAAD5.