h13080 Vnxe3200 Unified Snapshots WP
h13080 Vnxe3200 Unified Snapshots WP
Abstract
This white paper reviews and explains the various operations,
limitations, and best practices supported by the Unified
Snapshots feature on the VNXe3200 system.
February 2015
Copyright © 2015 EMC Corporation. All Rights Reserved.
For the most up-to-date listing of EMC product names, see EMC
Corporation Trademarks on EMC.com.
Audience
This white paper is intended for EMC customers, partners, and employees who want
to understand how to leverage Unified Snapshots on the VNXe3200 system.
Redirect on Write
Unified Snapshots uses Redirect on Write (ROW) technology. After a snapshot is
taken, new writes to the LUN, file system, or their snapshots are redirected and
written to a new location within the storage pool. New data is written to a new block
while the old data remains in place. This is more efficient than copy on write (COW) or
copy on first write (COFW) because no data needs to be moved from one place to
another.
LUN Groups
The VNXe3200 system introduces the concept of LUN Groups. A LUN Group is a
collection of one or more LUNs and is used to preserve consistency. The Unified
Snapshots operations supported for a LUN can also be executed on a LUN Group. LUN
Groups can be used to group together a set of LUNs that belong to the same
application.
With Unified Snapshots, LUN Groups are able to provide crash consistency. Unified
Snapshots provide the ability to take point-in-time snapshots of all the LUNs in a
group, instead of taking an individual snapshot of each LUN at varying points in time.
When the snapshot of a LUN Group is taken, all writes to the LUNs within that LUN
Group are held until their snapshots have been created.
The most common use case for LUN Groups is for maintaining crash consistency
among LUNs. For example, a database application which has a Data LUN and a Log
LUN can be placed in a LUN Group (Figure 2). In this example, the information in both
the Data and Log LUNs are dependent on each other and a LUN group can be used to
perform operations on the LUNs together.
Local Protection
Snapshots can be used for creating periodical point-in-time copies of data. Unified
Snapshots provide the ability to restore data from a point-in-time in the event of a
data loss or corruption of data which is being accessed by a host. Snapshots provide
local protection; data can be restored directly from the storage array. Snapshots can
also be used for backups by providing snapshot access to a backup server.
In Figure 3, a LUN is scheduled for snapshots to be taken daily. Snapshots are taken
on Monday, Tuesday, and Wednesday. If the data on the LUN is accidently deleted on
Snapshot Scheduling
Snapshot schedules for a LUN or file system can be assigned both while creating a
snapshot and after the LUN or file system has been created. During the LUN or file
system creation process there is a step named Configure Snapshot Schedule (Figure
5). The snapshot schedule may be chosen from a group of predefined and user-
created snapshot schedules. Also, new schedules can be created.
Snapshot Auto-Delete
Unified Snapshots provide the ability to automatically delete snapshots. The main
goal of Auto-Delete is to manage pool space automatically, so that snapshots do not
consume space when they are no longer needed. As snapshots consume usable
capacity from a storage pool, it is important to balance protection needs with user
space. Snapshot Auto-Delete can be configured to delete snapshots based on the
space utilized in the pool or by expiration time.
Delete Eligibility
Auto-Delete can be enabled for snapshots of: LUNs, LUN Groups, file systems, and
VMware datastores. When a storage pool or storage resource is being created, the
Auto-Delete settings can be configured when assigning a snapshot schedule to a
storage resource. The Auto-Delete settings can also be modified after a pool or
storage resource has been created.
To view or modify the auto-delete option, navigate to the details page of your storage
resource, click the Snapshot tab, select the snapshot to modify, and click Details. In
the Auto-Delete tab, you can review and modify the settings (Figure 8).
Figure 8: Auto-Delete
If a snapshot is configured with the default Auto-Delete option of Pool auto-delete
threshold setting (Figure 8), the settings of the pool are used to decide when Auto-
Delete is initiated for a snapshot (Figure 9). By default, only the total pool space
threshold option is enabled, and Auto-Delete state is idle, which means it is not
Auto-Delete Thresholds
Note: If a snapshot is attached it will not be deleted during the Auto-Delete operation.
For a snapshot to be subject to the Auto-Delete process, the Auto-Delete option must
be enabled at both the pool level and snapshot level (Table 1).
Table 1: Auto-Delete Behavior
Pool Snapshot Auto-Delete Behavior
ON ON Snapshot is subject to Auto-Delete
ON OFF Snapshot is exempt from Auto-Delete
OFF ON All snapshots in the pool are exempt from Auto-Delete
OFF OFF All snapshots in the pool are exempt from Auto-Delete
When the storage pool capacity exceeds the system defined pool threshold of 85%, a
pop-up appears in Unisphere (Figure 11). This is a warning alert to expand the storage
pool or delete any storage resources or snapshots which are no longer being used in
the storage pool.
Snapshot Expiration
Snapshots can be configured with an expiration value, which pre-determines when
they will be deleted. However, when the expiration value of a snapshot is reached,
the snapshot may not be destroyed immediately. The system scans for expired
snapshots once an hour and runs a deletion process which deletes eligible snapshots
during this scan. The name of a snapshot that is eligible for deletion is affixed with
Destroying, to show that the snapshot is being deleted.
Figure 15 shows that the expiration value for a snapshot can be set in the Auto-Delete
tab of the snapshot details. The minimum expiration value for a snapshot is 1 hour.
Once a snapshot is configured with an expiration value, the date and time when the
snapshot will expire is shown.
Creating Snapshots
Once a snapshot of a LUN is created, a point-in-time copy of the data of the LUN is
made. Immediately after a snapshot is created no space is allocated to the snapshot,
yet space will be allocated to the snapshot as new data is written. The space required
for the snapshot is allocated from the same storage pool from which the LUN was
created.
Snapshots can be created for both thick and thin LUNs. Once a snapshot of a LUN is
created, performance may be impacted on the newly written or snapped area of the
LUN. Depending on the workload, the first snapshot of a LUN may have a performance
impact, but performance will not be impacted as additional snapshots are created.
Performance will recover after the last snapshot is deleted.
To create a snapshot, select a block storage resource and click on Details. In the
Snapshots tab, you can you use Create Snapshot to create a snapshot of the storage
resource (Figure 16).
When creating a snap, you have to assign a name, a brief description (optional), and
configure the Auto-Delete policy. The default name for a snapshot is “Date_Time” (ex.
“2104-04-21_14.37.05”). The default option for the Auto-Delete policy is Pool auto-
delete threshold setting. Details of the Auto-Delete policy are explained in the
Snapshot Auto-Delete section.
Note: When a snapshot is initially created, the Attached and Modified values are set
to No. The snapshot becomes modified after it is attached to a host.
Note: After a snapshot is attached, the Attached and Modified states are changed to
Yes.
The detach option can be used to remove host access to a snapshot. When a
snapshot is detached from the host, the Attached State changes to No and the Last
Writable timestamp is updated to reflect the time when the snapshot was detached.
Storage is not returned to the pool after a snapshot is detached; it is only returned
after the snapshot is deleted.
To detach a snapshot, select a block storage resource and click on Details. In the
Snapshots tab, you can you use the Detach Snapshot option to detach a snapshot
from a host.
Note: Only one snapshot of a storage resource can be attached to a host. Hence,
snapshots which are not being used should be detached.
Restoring Snapshots
The snapshot restore operation provides the ability to revert a snapshot back to the
point-in-time when the snapshot was taken. The restore option is only available for a
LUN that is not currently attached to a host. For a LUN Group, the restore operation is
enacted on all LUN Group members.
To restore data from a snapshot, select a block storage resource and click on Details.
In the Snapshots tab, select a snapshot and use the Restore Snapshot option to
restore from the snapshot (Figure 20).
Creating Snapshots
Once a snapshot of a file system is created, it creates a point in time copy of data of
the file system. Immediately after a snapshot is created there is no space allocated to
the snapshot. Space will be allocated to the snapshot as new data is written to the
Deleting Snapshots
The snapshot delete operation for file storage resources works identically to the
delete operation on block storage resources. More details are provided in the
Deleting Snapshots section.
Limits
Table 2 displays the numerical limits for Unified Snapshots on the VNXe3200 system.
Table 2: Unified Snapshots Limits
Name Limit
Maximum LUN Snapshots per Array 1000
Maximum Snapshots per LUN 256
Maximum Block Snapshot Level 2
Maximum File System Snapshots per Array 250
Maximum Snapshots per File System 96
Maximum File Snapshot Level 10
Maximum User File Systems + Snapshots per Array 500
Best Practices
Best practices when using Unified Snapshots include:
It is recommended to use thin (default) LUNs and file systems if snapshots are
going to be enabled. Snapshot creation will have lower performance impact on a
thin storage resource as compared to a thick LUN or file system. Thin LUNs and file
systems also improve space efficiency.
Plan snapshot deletions to occur during non-peak times. If snapshots have to be
deleted during peak hours, reduce the number of concurrent snapshot deletions.
On a thick storage resource, do not delete the final snapshot if additional
snapshots will be taken in the future.
Place LUNs that depend on each other into a LUN Group in order to maintain crash
consistency. Perform snapshot operations on the LUN Group.
Auto-Delete policies should be configured on your storage resources to have the
system automatically delete snapshots which are no longer needed. This
mitigates the potential of reaching threshold limits that prevent new snapshots
Interoperability
Unified Snapshots is designed to work with the other VNXe3200 features. This
integration enables easy management and the ability to utilize the features with
snapshots created on the VNXe3200 system.
References
The following documentation can be found on EMC Online Support:
Introduction to the EMC VNXe3200
Introduction to the EMC VNXe3200 FAST Suite
Unisphere for the VNXe3200: Next-Generation Storage Management
Using a VNXe3200 System with CIFS File Systems
Using a VNXe3200 system with Fibre Channel or iSCSI LUNs
VNXe3200 Best Practices for Performance
VNXe3200 File Deduplication and Compression