Know your Volumes
XtremIO has just released a new feature. The feature name is DRR Per Volume. The DRR Per Volume Reporting feature allows users to calculate data savings for Volumes and Volume Snapshot Groups (VSGs). The data savings calculations will provide the following information for each Volume and/or Volume Snapshot Group (VSG):
- Unique Physical Space per Volume or VSG: The minimum capacity that will be freed if this Volume/VSG is deleted
- Data Reduction Rate (DRR) Per Volume or VSG: Estimated Data Reduction Ratio of the Volume/VSG, taking into account Deduplication and Compression
- Copy Efficiency per VSG: Estimated efficiency of using volume copies (rather than independent volumes)
The Volume Snapshot Group (VSG) Structure
Creating a Copy in XtremIO is instantaneous. This is because when a copy is created, no data or meta data is copied. Rather, a new internal volume construct is created. This internal construct stores all of the shared data between the Volume and it’s Copy.
In the example below, when a Copy is taken from the Product Volume, and internal construct is created and stores all of the shared data of the Production Volume and its Copy.
All the Copies taken from the original Volume are part of the same Volume Snapshot Group (VSG). This includes also copies taken of existing copies.
Unique Physical Space
The Unique Physical Space calculation takes all the address space of a Volume and calculates how many of its logical blocks reference a physical block with a single logical address reference (reference count =1). This means that if this logical block will be deleted, than it will definitely free up a physical block too.
What does Global Deduplication has to do with Unique Physical Space Calculation?
The Unique Physical Space is the minimum capacity that will be freed if this Volume/VSG is deleted. You may ask, why is this this the minimum unique physical space that will be freed and not the actual? Well this relates to how XtremIO calculates Deduplication Globally.
There are 2 possible ways to calculate Deduplication in storage products:
- Volume based Deduplication – In this method, only duplicated blocks within a volume are deduped. If the same data block is written to two different volumes, it consumes two physical blocks.
- Global Deduplication – In this method, all duplicated blocks within a Cluster are deduped. If the same data block is written to two different volumes, it will consume only a single physical block.
In order to promote maximum efficiencies, XtremIO architecture is based on an in line global deduplication data service. A duplicated block will always be written only ONCE in the XtremIO cluster.
The Unique Physical Space counter counts all blocks that have a single occurrence (reference count) in the array – meaning all blocks that are not deduped. XtremIO doesn’t differentiate between blocks that are unique to a volume and blocks that are shared between different volumes. It only counts completely unique blocks. If a block has only 2 copies, and both copies are stored on a given Volume, the Unique Physical Space will not include this block in its calculation, even though it will be removed if the Volume is removed.
Data Reduction Rate (DRR)
The DRR Per Volume and VSG is calculated by taking the data blocks of the Volume/VSG and calculating their Deduplication and Compression rates. In most cases, Volumes in the same VSG will have similar DRR rates.
In the example below, the DRR rate of a VDI volume is 13.4:1. The DRR rate of an Oracle volume is 3.2:1.
Copy Efficiency per VSG
The Copy Efficiency calculates per VSG the ratio between all the host accessible blocks of the Volumes of the VSG and the amount of logical blocks that are actually stored in the VSG. The relative amount of shared blocks affects the Copy Efficiency ratio. The more blocks that are shared within the VSG, the higher the Copy Efficiency.
Let’s review the following example. The below VSG has 3 external Volumes: Production Volume, Writable Copy1 and Writable Copy2.
The number of logical blocks stored in the VSG is: 18
|Volume||Logical blocks stored in VSG|
|Total logical blocks||18|
However, the volumes and writable copies has the following host accessible address space:
|Volume||Host accessible address space|
|Host accessible blocks of VSG||36|
Therefore, the Copy Efficiency of the VSG will be: 36:18 which is a 2:1 ratio.
Running the Calculations
The data saving calculation are performed at the Volume and Volume Snapshot Group (VSG) levels. User can provide a Consistency Group as input, and the savings metrics will be automatically calculated for all of the Volumes of the Consistency Group. The calculation is always made on all of the Volumes of the selected Volume’s Snapshot Group.
scanning a volume and identifying unique blocks is a resource and time-consuming task, the array will only take samples of the volume and analyze them.