Cloud-based backup holds broad appeal for many businesses because it offers a low-cost way to protect business
data off-site. Beware, however. Moving business data into the cloud is the easy part. Getting it back when you really need it is when things can get challenging. For this reason, IT planners need to understand all the caveats before they embrace cloud-based data protection as an integral part of their recovery strategy.
First, consider how most backup as a service offerings migrate data between the end-user location and the cloud service provider's (CSP's) data center. To economize on bandwidth and speed up transfer times over WAN or Internet connections, backup data typically is deduplicated and compressed before it's transmitted to the CSP facility. This is a great way to efficiently move data off-site. However, it presents some significant challenges when data recovery needs to take place.
First, deduplicated backup data must be "rehydrated," or reconstituted into its native format before it can be recovered to an application. The rehydration process itself takes some time to occur, but the bigger challenge is the time it will take to retransmit all this data from the CSP site back to the customer location. This is generally not a major issue if a single file needs to be recovered. However, if a large application instance -- or worse, an entire corporate application environment -- needs to be restored, the time it will take to trickle the data over network links could translate into multiple days, if not weeks.
To get around this problem, an increasing number of CSPs are starting to offer disaster recovery as a service (DRaaS), whereby the subscriber can run their business application as a virtual machine (VM) instance in the CSP's cloud. While this mitigates the need to return an application to operation in the customer data center quickly, it does present unique challenges.
First, will the application even run adequately in the provider's cloud? For example, to run at production service-level quality, the application might need high-speed storage resources and/or need to be attached to a robust network that can support hundreds or thousands of simultaneous user sessions. Will equivalent resources be available in the cloud if you need them? In short, moving an application from an environment that you know and control to an environment where you are totally dependent on the CSP may introduce some unknown variables that could negatively impact quality of service.
In fact, most DRaaS providers don't provide any service-level agreements (SLAs) around the performance of an application recovered into their cloud. Many merely enable you to run the application in their facility.
Moreover, when the time comes to move the application back to your data center, how will you accomplish that? All the data will still have to transfer over the network to your site.
Given these challenges, what is the best option? Obviously, if the application can tolerate limited service quality for a short time until normal service can be restored within your data center, then recovering in the cloud could be a viable option. It is critically important, however, that you first work with your cloud provider on a plan to facilitate bulk data transfers from their facility to your data center to enable the rapid recovery of your information.
Another scenario that businesses need to consider is what happens if dozens or hundreds of the CSP's clients declare a disaster at the same time? Resource contention within the CSP data center could bring recovery operations to an absolute standstill. While many CSPs may attach SLAs to recovery times, there is virtually nothing they can do to ensure that all their clients will have speedy access to their backup data during a regional disaster.
Does this mean that performing recovery from data in the cloud holds no value to businesses? Absolutely not. All it means is that IT planners need to devise ways to plug the holes in their recovery strategy and try to prepare the best they can for all contingencies. For example, maintaining a copy of backup data to tape and storing it locally -- or ideally, in a secured facility where it can be retrieved within a 12- to 24-hour time frame -- would actually be a better recovery option when data has to be restored en masse.
For more on this topic
Cloud-based backup and software deduplication explained
Data backup software and cloud-based backup working together
Evaluating disk- and cloud-based backup options
In other words, a local, tape-based recovery is going to perform much faster than a recovery that has to trickle over WAN or Internet links, particularly when those links are saturated with other user traffic. Having a tape copy also provides an additional layer of protection in case backup data replicated into the cloud gets corrupted somehow.
At first glance, it might seem that having a tape-based backup strategy defeats the purpose of going to cloud-based backup. After all, eliminating tape holds strong appeal for many organizations.
The good news is that when cloud-based backup is used in tandem with local tape backups, organizations can actually have the best of both worlds -- the convenience of conducting a cloud-based recovery where it presents less risk -- such as with single-file restore jobs -- while having the security of local tape to conduct more rigorous recovery tasks, such as full system recoveries. Plus, since a copy of the backup data resides in the cloud, it would be necessary to cut only one copy of the backup to tape, rather than maintain two copies (one for on-site and one for off-site). For many organizations, this would be a big upgrade to their backup operational and recovery strategy.
Backup and recovery in the cloud is one tool in the data protection arsenal. It is not a one-size-fits-all solution that will solve all your backup and recovery needs, but it is an excellent option for enabling organizations to secure and protect data in an alternate facility while providing limited recoverability options. When combined with a local data recovery strategy, like cost-efficient tape backups, IT organizations can enhance business data protection without breaking the bank.
Likewise, DRaaS offerings that enable organizations to recover an application as a VM in the cloud should be vetted before they're adopted by end users. If an end user simply wants to be able to run an application, even in a degraded mode, while recovery operations are taking place in the primary data center, a DRaaS product could be an acceptable cost-effective alternative to a dedicated DR infrastructure.