In my first blog I was talking about the XtremIO snapshots . In this blog I would like to delve more into the XtremIO snapshot features which offer a large number of features like high performance, low latency, read/writeable snapshots etc.
Snapshots are efficient in metadata and physical space can be created instantaneously ,have no performance impact and have the same data services as any other volume in the cluster (such as thin provisioning and Inline Data Reduction).XtremIO snapshots are easy to use and manage and leverage a sophisticated metadata management engine that provides superior support for flash media enabling high performance snapshotting. XtremIO snapshots can be used for Development and Test, backups, protection against logical corruption, offload processing (real-time analytics, SAP landscape, and more) and consolidation of testing and development with production.
XtremIO Snapshots render the following benefits which are listed below:-
- Snapshots are created instantaneously and provide a workable copy of the production volume.
- Snapshots are read-write copies of source volumes.
- Snapshots appear and are managed in the cluster as regular volumes.
- Snapshots have the same data services as any volume in the system; the inline global deduplication and thin provisioning features are in constant operation.
- XtremIO Snapshots are both metadata and data efficient.
- Snapshotting requires no reserved space.
- The system supports consistent snapshots on multiple volumes.
- Snapshots can be on snapshots at any hierarchy or span.
- Deletion of a volume or any of its snapshots does not affect the child snapshot or parent snapshot/volume.
- The system provides predictable and consistent performance on production volumes and snapshots.
The comparison between the features and benefits of XtremIO vis-à-vis different snapshot technologies has been depicted in the below table:-
From architectural standpoint , XtremIO provides the following advantages :-
- Space efficient metadata
- Creation of a snapshot does not consume metadata.
- Metadata is consumed only for new and globally unique data blocks
- No performance impact
- There is equal performance with snapshots and volumes.
- No impact occurs when snapshots are created.
- Read performance is equal on all snapshot hierarchy levels.
- Inline-Data-Reduction-enhanced "redirect-on-unique-write" method
- Consumes space only for new unique data blocks
- There is no physical data movement on new writes.
- Even distribution of system resources
- All Storage Controllers in the cluster are constantly engaged in managing the I/O data flow and metadata, regardless of the entity type.
- More CPU power and memory remain available (a single controller, versus multiple controllers).
- There is consistent even distribution of workload across all available resources.
- Optimized for flash memory
- XtremIO snapshots are optimized for maximum flash endurance.
- No data movements occur on snapshot creation or during writes.
- Inline Data Reduction provides added value in capacity efficiency and flash endurance.
- There are additional benefits with the flash media (in terms of performance).
- The system provides efficient metadata and capacity utilization.
XtremIO's snapshot technology is implemented by leveraging the array's content addressing capabilities along with in-memory metadata and the system's dual stage metadata (which results in Inline Data Reduction), optimized for SSD media with a unique metadata tree structure that directs I/Os to the right data timestamp. This allows efficient snapshotting that sustains high performance, while maximizing media endurance, both in terms of ability to create multiple snapshots and the amount of I/Os that a snapshot can support.
In the logical volume diagram shown in the above figure , every written block address is mapped to a fingerprint. Therefore, this mapping is called Address-to-Content. Additionally, there is a separate metadata mapping from the content to the actual unique physical blocks that are written to the SSDs (thus forming the dual-stage metadata structure).
As every XtremIO volume is thinly provisioned, the addresses that have not been written remain empty and do not occupy metadata (or data) space. Therefore, the XtremIO thin provisioning is 100% space efficient.
When creating a snapshot, the system generates a pointer to the ancestor metadata of the actual volume in the cluster. Therefore, creating a snapshot is a very quick operation that does not have any impact on the cluster and does not consume any physical or logical capacity. Snapshot capacity consumption occurs only if a change requires writing a new block.
When a snapshot is created, the existing metadata for the volume becomes an "ancestor" entity that is shared between the production volume and the snapshot. New empty containers are created for subsequent changes to the production volume and the snapshot volume. Thus, the act of creating a snapshot is instantaneous and involves no data or metadata copies.
In the next blog , I am going to discuss some more good features and benefits of XtremIO Snapshots.
- For getting details regarding XtremIO array, pls. visit the site here.
- Pls. visit the EMC Store for the datasheet of XtremIO array here.
Socialize this blog using bitly: http://bit.ly/W70Q0i
A deep dive into #XtremIO snapshots for #Oracle in this new blog: Features and Benefits of using XtremIO: http://bit.ly/W70Q0i #EMC