One of the key challenges to any data protection strategy is deciding where to store data and how to get that data off-site. Initially in most environments, all backups were stored on tape. Those tapes were then packaged up and shipped to an off-site location. Then disk-based backup appliances came to market and, leveraging deduplication, were able to replicate data directly to a remote location. Each of these had challenges and the cloud is being viewed as a way to overcome those challenges. But the cloud also has problems, especially in terms of recovery. As a result, a mix or hybrid approach is now being offered as the nirvana of data protection.
The Problems With Traditional Data Protection
Traditional backup storage, tape and disk, had limitations when it came to getting data off-site. First, both required that a remote location was available to the organization. This of course is not always the case. Many organizations don’t have multiple offices, or if they do, those offices are not equipped with appropriate IT resources to be on the receiving end of the data protection workload.
Even when just on-premise, these systems can create problems for the IT administrator. Tape, while offering cost effective scaling, can be difficult to manage day to day. Also, the purchase of a library can be expensive for many small to medium sized business. To be able to backup at the full potential of tape requires a significant investment in backup network architecture. Tape, because of its portability, is also susceptible to drops and mishandling, which may eventually lead to corruption. When it comes to recovery, especially of smaller data sets, it can be time consuming. Finally, tape is not suitable for any of the newer rapid recovery features like “recovery-in-place”.
Disk, especially for the small to medium sized business, is more ideal. It does not need to be streamed like tape does, so a non-optimal network won’t adversely impact performance. And features like deduplication and compression have made the use of disk much more affordable. In terms of recovery, the size of the data being recovered has little impact on the recovery time. Again, thanks to deduplication and compression, transfer to a remote site, if available, is more automated. Finally, new features like changed block backups, recovery-in-place, and changed block recovery are all supported by disk.
While disk is ideal for the most recent copies of backup, it can be difficult and expensive to scale for the long term retention of data. The cost to power and cool as well as allocate data center floor space to the disk backup appliance can end up being a sizable expense. And there is the issue of making sure that there is a secondary site available with IT personnel in the event of a disaster.
The Problems With Traditional Cloud
The cloud promises to solve many of the challenges of disk-based backup technologies. First, it by definition provides a secondary site with IT staff. Second, the cost of buying and maintaining an on-site storage system is eliminated. Finally, many cloud backup providers have evolved into Disaster Recovery as a Service (DRaaS) providers that are focused on significantly reducing recovery time in the event of a disaster.
But the cloud is not perfect. The key issue with the cloud is its inherent latency, the time it takes to transfer data from the cloud back to the primary data center. While DRaaS does resolve some of this transfer time, most IT professionals would rather recover to their own data center first, if it is available, and only use a cloud recovery as a recovery of last resort. Even in that last resort circumstance, there is going to come a time in which the application has to move from the cloud provider’s data center back to the primary data center. While some of the data can be pre-staged, there is always going to be some data that has to be transferred over the internet connection as a final sync-up. While that transfer occurs, the application will likely be down.
What is Hybrid Cloud Backup?
Hybrid cloud backup promises to resolve the latency issues of a cloud-only backup and the scaling issues of an on-premise disk-based backup appliance strategy. Typically a hybrid cloud backup solution consists of an on-premise appliance that has enough capacity to hold several full backups, and the incremental backups that would be created between those fulls. The hybrid appliance becomes the point of first restore, since it is disk based and on-site.
The key difference between disk backup and hybrid cloud backup is what happens after the backup is complete. The hybrid systems add an extra step and replicates backed up data directly to the provider’s cloud or to a public cloud provider like Amazon AWS or Azure. This has several advantages over traditional disk-based backup.
First of all, the on-site appliance in the cloud strategy can be far simpler and less expensive, as it does not need the level of hardware complexity to provide resiliency from failure that disk appliance have, since data could be pulled from the second copy in the cloud if the hybrid appliance were to fail. Also, replication pre-seeds the DR data in the provider’s facility. Third, the hybrid appliance grows much more slowly than a disk appliance would, since it only needs to be large enough to store a few backups and the associated incremental data. Finally, most hybrid cloud appliances also come with all the software required to do the actual backups. Most disk appliances require the purchase of a secondary backup software system.
In terms of recovery the hybrid appliance becomes the primary recovery point, being used for rapid recovery of corrupted applications. Since most recoveries are of data from the most recent backup, it stands to reason that most recoveries would come from the hybrid appliance. The advantage with this design is that the large majority of recoveries are happening at a data center over LAN speeds instead of across a WAN connection to the cloud provider.
There are only two instances where recovery would come from the cloud. The first is an archive recovery, a recovery of data that is not available on the local appliance. In most cases, there is no pressure to restore this data rapidly, even a few hours is more than acceptable.
The other instance is the very rare case of a disaster. While rare, it is of course critical that the recovery effort works and does so quickly. Again, this is a potential weak spot for cloud backup. But with DRaaS, the applications can be started in the provider’s cloud while the primary data center is returned to operations.
The hybrid appliance plays an important role in DRaaS. First, if the appliance can survive the disaster then only the data that was changed or added while the application was executing in the provider’s cloud needs to be recovered across the internet connection. This could save a tremendous amount of time. Second, even if the original appliance is lost, a new appliance can be seeded at the cloud provider’s facility and shipped to the primary data center, again reducing downtime.
Another key consideration for prospective hybrid cloud customers is the flexibility of the cloud destination. Ideally the customer should have choice to store their data in a public provider facility or a facility provided by the vendor. Further, they should also provide the ability for the customer to stand up their own private cloud destination as the customer’s own organization begins to scale.
Conclusion: The State of Hybrid Cloud Backup
The challenge with hybrid cloud backup is the number of potential solutions in the space. Each varies in its level of providing complete protection. MSPs and their customers should be on the lookout for a solution that can deliver both physical server data protection as well as virtual server protection. It is also reasonable to expect that the solution provide endpoint protection and, of course, DRaaS is a must have.
These capabilities, while rare to find in a single solution, are just the beginning. For example, companies like Infrascale can provide all of the above plus endpoint protection, collaboration and file sharing and this is a trend we expect to continue. Once you allow a company to manage all of your data in their cloud, there is almost no end to the number of additional services they can provide to you.
Sponsored by Infrascale