luchschen_shutter - Fotolia
Published: 03 Oct 2016
Until relatively recently, hardly anyone liked performing backups. IT professionals commonly regarded the technology...
as stale and stagnant, while their bosses largely thought of backups as akin to insurance -- a necessary expense with no ROI unless a disaster occured.
Rapid technological innovation has led to a backup renaissance over the last several years, however, with modern backup products delivering previously unimaginable capabilities. Some of these new features even provide business value, no disaster required. This innovation in backup technology is largely attributable to two trends: the ever-increasing impracticality of the backup window and the adoption of server virtualization.
The end of backup stagnation
Legacy scheduled backups are, by nature, disruptive to other business processes. For many organizations, there is literally never a good time to run a backup, and even if they have the luxury of scheduling nightly backups without concern of disruption, the backup window itself will eventually become inadequate.
Most enterprises accumulate data on a daily basis, and there is only so much data that can be backed up within a finite backup window. Although techniques such as incremental forever backups and global deduplication limit the amount of data to back up each night, it's usually only a matter of time before the volume of data to back up exceeds the backup window.
Backup vendors eliminated the problems associated with backup windows through the development of continuous data protection, which creates very frequent, block-level backups rather than monolithic nightly backups. So, every few minutes, any newly created or modified storage blocks are copied to the backup target -- usually a storage array -- completely eliminating the need for a backup window.
As an added bonus, continuous data protection greatly reduces the recovery point objective (RPO), which is approximately 24 hours for legacy nightly backups. An RPO of that length could cause an organization to lose nearly a full day's worth of data due to the infrequency of its backups. Continuous data protection, by contrast, could limit potential data loss to less than five minutes.
Server virtualization is the second trend largely responsible for the end of backup stagnation. Virtual data centers allow you to protect and recover data in ways difficult or impossible to achieve in physical server environments. In fact, it is server virtualization that makes instant recovery technology possible.
The convergence of backup and disaster recovery is one of the major trends in data protection over the last few years. Backups exist as point-in-time copies of systems or data that can be restored on an as-needed basis. Disaster recovery, on the other hand, has always focused on continuity of business. It does not seek to restore data following a data-loss event, but rather to keep mission-critical systems running in the event of an outage.
Instant recovery converges backup and DR by offering the best of both worlds.
Instant recovery sandboxes
Instant recovery enables organizations to mount and use a backup copy of a virtual machine. This allows the VM to be brought back online within a matter of a few minutes, as opposed to the far lengthier amount of time it would take to restore from a traditional backup. Although the primary goal of instant recovery is to dramatically shorten the VM RTO, some vendors have added the ability to create sandboxed environments using the same technology.
For example, Veeam leverages the same mechanism used for the instant recovery of VMs to offer a virtual lab feature that perfectly mimics production environments. Because these virtual labs run from the backup server, the sandboxed VMs don't require significantly more storage space beyond what's already in use. Virtual labs are useful for testing patches, operating system upgrades and new applications.
Veeam isn't the only vendor extending instant recovery to create sandboxed environments. Cohesity, for instance, offers a VM cloning feature that enables the creation of development and test environments through the use of copy data management.
One major problem that's plagued traditional backups is long recovery time objectives (RTOs). In other words, it takes time to restore a backup. Long recovery times are unacceptable for mission-critical systems. Instant recovery leverages the power of server virtualization to shorten recovery time to just a few minutes.
Instant recovery is based on the idea that if you use disk-based backups, as nearly all continuous data protection systems do, then a full copy of any protected virtual machine (VM) resides on the backup's storage array. Rather than waiting for a restoration to complete, a VM can be mounted directly from the backup storage array and immediately brought into service. Of course, the problem with using backup copies of VMs for production is it causes backups to be modified. Instant recovery protects backup copies of VMs against unwanted modifications by using hypervisor snapshots.
Vendors originally introduced hypervisor snapshots as a convenience feature to allow admins to roll back a VM to a previous state without the need to restore a backup. When an administrator takes a snapshot of a VM, the hypervisor creates a differencing disk. All of the VM's write operations are redirected to this differencing disk, which means that the VM's original virtual hard disk is left unaltered.
Backup products with instant recovery capabilities take advantage of hypervisor snapshots by creating a snapshot of a VM before bringing a backup copy of the VM online in an unaltered state. Once the VM is live, a more traditional restoration operation is initiated. This operation restores the VM to its original location or to an alternate location. When the restoration completes, the contents of the differencing disk are merged into the newly restored VM, which is then put into service. At that point, the backup VM copy is dismounted, the differencing disk discarded and backup operations resume normal operations.
Most enterprise backup vendors offer an instant recovery feature for virtual data centers. These capabilities tend to be very similar in scope from one vendor's offering to the next. The most common differences are usually related to hypervisor support. Veritas NetBackup, for example, provides instant recovery capabilities for VMware, while Acronis and Veeam enable instant recovery for both VMware and Hyper-V.
Flat backups, another type of backup gaining popularity, are based around the creation and replication of storage snapshots. Although functionally very similar to hypervisor snapshots, storage snapshots are created at the storage array level rather than the hypervisor level.
Because flat backups occur at the storage level, they must be supported by the storage vendor. Storage snapshots are vendor-proprietary, and not every storage device includes native snapshot capabilities. As such, flat backups aren't an option for every organization. This is especially true for those with heterogeneous storage hardware. Even so, enterprise storage vendors such as EMC and Hewlett Packard Enterprise (HPE) support flat backups.
Just as technological advances have made backups better, these very same innovations have also led to improved backup testing. For many years, backup testing was largely synonymous with verification, where you compared backups against original data to ensure it was backed up correctly. However, the simple verification of data doesn't guarantee the integrity of the backup.
Suppose for a moment that the Winload file is deleted from a Windows server. The server would continue to function normally because that file is only used during the boot process. If the server is backed up, a verification of the backup would succeed because the backup contents match the server's hard disk. However, should you restore the backup, the restored system would be unbootable because a critical system file is missing from the backup.
Veeam goes beyond basic backup verification to perform a number of predefined backup recovery verification tests, where VM backups are booted into a sandboxed environment and then tested. Some of the automated tests include heartbeat, ping, application testing and backup file validation. These tests allow administrators to verify backups at a functional level, not just at the file level.
Alternative backup methods such as continuous data protection remain more popular than flat backups. This is largely attributable to flat backup hardware requirements and the proprietary nature and high cost of such hardware. Even so, flat backups are acquiring mainstream acceptance thanks largely to the introduction of application consistent storage snapshots. Until relatively recently, storage snapshots were designed to be crash-, but not application-consistent, which means they weren't an appropriate choice for protecting application servers. Today, vendors such as EMC, Hitachi Data Systems and HPE allow for the creation of application-specific storage snapshots.
Flat backups will likely become even more popular as they get less proprietary. Veeam, for instance, supports backup and replication of storage snapshots for a variety of storage hardware through a single backup application. The fact that storage hardware remains proprietary means Veeam's product isn't truly universal. Veeam is working to add support for more storage hardware, however.
Another major backup trend in today's backup renaissance is the widespread adoption of backup appliances. While backup appliances vary considerably in both scope and capability, they are (generally speaking) dedicated storage devices that act as backup targets -- often by emulating a tape library.
Since backup appliances are usually disk-based storage devices, it's tempting to think of them as hardware. However, although vendors such as Unitrends and Veritas offer physical backup appliances, virtual appliances also exist. Amazon Web Services, for example, allows customers to back up resources to the cloud by leveraging a virtual backup appliance.
Backup appliances, meanwhile, also vary widely with regard to functionality. Veritas designs its appliance to function as a master server, a media server or both, for example. These appliances include features such as snapshot replication, dynamic storage and WAN acceleration.
Unitrends takes a slightly different approach by offering a hyper-converged, all-in-one backup appliance designed to provide instant recovery capabilities. Its Recovery Series appliances also include features such as adaptive deduplication, local archiving and WAN optimization.
Backups have evolved to provide capabilities once considered impossible. Better still, backup innovation shows no sign of slowing down. For example, future backup products will likely emphasize seamless backups of a wide variety of cloud-based resources. What else is in store for these once maligned yet now vibrant and essential processes and technology? I'm sure we will soon find out.
About the author:
Brien Posey is a Microsoft MVP with two decades of IT experience. Previously, Brien was CIO for a national chain of hospitals and health care facilities.
Data backup technology pushes appliance advances
Backup technology, snapshots come together
Today's data backup options