alunablue - stock.adobe.com
Snapshots 101: Array vs backup software
We run the rule over snapshots in storage arrays and backup software and find a range of capabilities that go from mere support to fully snapshot-centric approaches
Nearly all enterprise and mid-range storage arrays come with snapshots. But that’s not the only place you can manage them from.
Backup software also comes with snapshot functionality, and in fact some of the more recent entrants to the market have built their approach to backup around snapshots.
Snapshots capability in backup products starts with the ability to manage and copy storage hardware makers’ snapshots. This is the case with Veritas, IBM’s Spectrum Protect and EMC’s Networker. Things get more sophisticated with, for example, Commvault, which can manage a range of array makers’ snapshots from a single console.
Then there is the more recent wave of data protection products that come as appliances and are somewhat akin to hyper-converged nodes. These products – such as those from Cohesity and Rubrik – could be said to be snapshot-centric and base their backups around them. Veeam also sits somewhere near these suppliers and their approaches, but without the hardware form factor.
Before looking at backup suppliers’ snapshot capabilities, let’s run over some snapshots fundamentals.
Snapshots basics
A snapshot is a copy of the state of a unit of storage (drive, volume, etc) at a specific point in time.
To be more accurate, snapshots are often comprised of pointers to an existing copy or copies – an original and/or subsequent snapshots – as well as actual storage blocks, such as those that have been deleted and need to be kept to rebuild the full picture at a specific point.
Storage arrays usually keep a set amount of snapshots (256 or 1,024, for example) and customers can roll back to a previous version if required. Depending on how often they have been made, that could give several days’ worth of coverage to restore to.
Where is it best to manage snapshots from – the array or your backup product?
Snapshots taken at the array are likely to be quick to take, with a lower input/output (I/O) tail as they are written locally to the same storage.
They will also be quick to access for the same reasons.
But snapshots on the array could become inaccessible if the array is effectively a single point of failure. Array-native snapshots are also likely to be crash-consistent, with a possibility that they lack essential components to allow for a full rehydration.
Snapshots made from backup software are likely to have a higher overhead on resources as they are managed from and written to other locations on the network.
They are also likely to take slightly longer and have more impact on production because they will be application-consistent. In other words, the snapshot will ensure the application is quiesced and everything needed to successfully rebuild data to that point in time is taken.
Array makers’ snapshot functionality
All arrays come with snapshots and, in general, they offer a similar set of choices.
Customers can set how many shots they want to retain (up to certain limits), whether they want them to be copy-on-write or redirect-on-write and in some cases (such as NetApp) whether they want regular snapshots, archive copies or clones that can be used for disaster recovery (DR) failover.
Snapshots in backup products
Cohesity
Cohesity’s approach is based on its DataPlatform and SpanFS file system, with snapshots and its SnapTree functionality operating as a metadata-based functionality within it.
In SnapTree, where normally recovery from snapshots would require the rebuilding of a large chain, Cohesity shifts some of that onto a chain of metadata pointers to gain a claimed boost in recovery times.
It calls SnapTree “distributed redirect-on-write”. Changes are written to new blocks every time and the “distributed” part refers to the fact changes are written across the file system.
Cohesity has native snapshot management integration with Pure Storage arrays, as well as HPE Nimble flash storage and Cisco HyperFlex NAS.
Commvault
In addition to traditional backup, Commvault allows customers to take backups of array snapshots. These can be stored elsewhere in case of problems on the storage array that cut access to snapshots held there.
Also, Commvault has IntelliSnap, which allows customers to manage multiple supplier array- and cloud-based snapshots through a single console and claims it can be used to base data protection on snapshots rather than traditional backup.
IntelliSnap can discover snapshot capabilities across the storage estate and cloud, and create long-term snapshot backups and archives from snapshots using commodity disk, cloud or tape.
EMC
EMC offers Avamar and NetWorker, both of which will work with VMware snapshots – as will most mainstream backup products – but of the two, the enterprise-level NetWorker has the more developed snapshot management capability. Its Snapshot Management automates the generation of point-in-time data snapshots and cloning on supported storage arrays such as EMC VNX, XtremIO, and Symmetrix.
NetWorker’s snapshot capability includes its module for Microsoft applications that uses Volume Shadow Copy Service (VSS) to protect Exchange, SQL Server, Active Directory, SharePoint and Hyper-V. Then there are its PowerSnap Modules for SAP, Oracle, SQL Server and IBM DB2 applications and databases.
NetWorker also has what it calls “snapshot-assisted backups” although these appear to be snapshots that are stored elsewhere for added protection or, for example, disaster recovery purposes.
IBM
IBM’s Spectrum Protect (formerly Tivoli Storage Manager, or TSM) features Spectrum Protect Snapshot, which allows backups to be created from a range of storage array makers’ snapshots, as well as those from database and applications such as enterprise resource planning (ERP) software. But it appears to need IBM storage hardware or a General Parallel File System (GPFS) as a target.
Rubrik
Snapshots are at the core of Rubrik’s approach to backup and data protection, using them to create a stable system image to make backups from.
In VMware environments it uses VMware’s vSphere APIs (application programming interfaces) for Data Protection (VADP) to create a snapshot of virtual machines (VMs) from which it makes backups.
For Windows environments, Rubrik wrote its own VSS agent for application-consistent snapshots. Writing its own VSS agent helps gives a claimed boost in immunisation against failed snapshots.
Also, Rubrik leverages snapshots in NAS backup where it works with NetApp OnTap snapshot and differential information to know what new blocks/files to backup since the previous job.
Veeam
Veeam has a range of functionality that aims to make storage array makers’ snapshots easier to work with. This includes the ability to copy storage snapshots to other locations and to restore items from it at fine levels of granularity.
You can configure a backup job to maintain a snapshot chain on the storage system in addition to regular backup files.
It is also possible to create snapshot-only jobs to build a chain of snapshots on the primary storage array and on secondary storage as an option.
Additionally, you can set preferences for numbers of snapshots retained and automatic deletion beyond set limits.
Meanwhile, Veeam’s Explorer for Storage Snapshots has integrations with Dell EMC, NetApp, HPE and IBM products.
Veeam also has Snapshot Hunter, which can detect and remove orphaned snapshots that remain after backup or replication job sessions to save space.
Veritas
In NetBackup, Replication Director uses the Veritas OpenStorage API to manage snapshots from hardware arrays and use them as roll-back points. Snapshots can also be replicated to other storage locations.
Meanwhile, Veritas’s multicloud CloudPoint product integrates with NetBackup and includes snapshot management and orchestration for storage array-based protection and in the cloud.
CloudPoint can connect to array-based snapshots in Hitachi Data Systems G-Series, HP 3PAR, Pure Storage, Dell EMC Unity and NetApp FAS arrays.
Incidentally, NetBackup uses its Snapshot Client to create an image of a client volume, then backs up data from the snapshot. According to Veritas, that means users can access primary data without interruption while data on the snapshot volume is backed up.
Read more about snapshots
- Cloud snapshots and backups: You often need more data protection than native cloud services will give. But should you choose backup or snapshots? And what about third-party backup in the cloud?
- Storage 101: Snapshots vs backup: We go over the basics of snapshots – they’re a quick and accessible way of protecting data, but they’re not a substitute for backup. So how do you combine the two?
