How cloud and virtualization are changing disaster recovery

It is impossible to have an IT modernization discussion that doesn’t include the cloud, but the details beyond that will vary greatly:

  • Some will replace their legacy backup product with a backup service.
  • Some should supplement their current backup product with cloud storage.
  • Some ought to add a cloud-based DR capability to their on-premises backup product.
  • Some are wrestling with how to protect the primary production workloads that have moved to the cloud.

Here’s a look at each of these four cloud and virtualization issues that impact data protection and disaster recovery in more detail, starting with replacing legacy backups with a backup service. Then, we’ll take a closer look at augmenting your existing backup product with cloud storage. Because some of you will choose to add a cloud-based disaster recoverycapability to your on-premises backup product, we’ll look at some things that merit consideration. Finally, we’ll investigate how you can protect your chief production workloads that have moved to the cloud.

Considerations for backup as a service

For organizations that are struggling with their legacy backup product, where any kind of upgrade is likely to be a significant replacement, backup as a service (BaaS) may be a good option. BaaS products enable a fresh start for data backup that changes the architecture of the backup product, the agent technologies and the economic model through with backups are achieved. BaaS products can also provide a different kind of agility, because the data is natively accessible or restorable from the cloud provider.

To be clear, BaaS is just like many other “as a service” offerings, in that they are a cloud‑based delivery of an IT function. The economics are different, the management experience is different, and the underlying infrastructure is designed to be delivered at enterprise-class scale by service providers, instead of a centralized IT department. But at its core, it is still just another backup product, with agent technologies on production platforms, backup schedules and restore jobs. As such, BaaS won’t fix infrastructure issues or unwieldy production servers that are hard to back up, or drastically change the administration time devoted to backup jobs or restore requests. This can be challenging when it comes to DR in the cloud.

Taking your current backup and recovery product to the cloud

For organizations whose current backup and recovery product has the modern platform capabilities that the organization needs and is performing at least adequately, BaaS may not be the best answer. Instead, most contemporary backup products have the ability to leverage cloud-based storage, as a supplement to the on-premises deployment. In so doing:

  • The existing backup agents on the production servers remain unchanged.
  • The existing backup jobs and schedules continue to operate unaffected.
  • The existing ability to recover quickly from local storage doesn’t diminish.
  • A tertiary copy of the backup data now resides in a cloud repository for compliance or DR purposes.

While all of that sounds good and easy, it can come with tradeoffs in that the means by which the data is replicated from the backup server to the cloud repository can vary greatly and will dramatically affect the agility and recovery options from the cloud copy. And, again, when it comes to cloud-based disaster recovery, you’ll want to consider both the pluses and minuses.

Most organizations should plan on a hybrid or D2D2C architecture

It is extremely difficult if not impossible for organizations of most sizes to maintain the service-level agreements that users and business owners have come to expect from recovery times and backup performance. Because of this, it is strongly recommended that most cloud-enabled backup and recovery products be “D2D2C” configurations — from production disk to local backup disks (D2D) before going to the cloud (2C). That said, D2D2C can take several permutations:

  • BaaS products with an intermediate caching product on-site before going to the BaaS repositories.
  • On-premises backup hardware replicating to a similar storage array at a service provider.
  • On-premises backup software replicating to another software instance at a service provider.
  • On-premises backup software writing to a cloud storage repository as a tertiary tier.

The method of replication and the type of cloud repository will directly affect the immediate usability of the cloud copy of the data, but some enable easier extensibility of existing backup software and/or hardware.

Cloud is not likely a tape killer. While other innovative IT technologies are usurping some usage of tape, one should not necessarily assume that D2D2C is an adequate replacement for D2D2T (tape), primarily due to most cloud providers’ inability or unwillingness to retain data for five, 10 or 15 years. Most cloud providers utilize disk as their repository, and therefore don’t have a cost-effective way to store data for that length of time.

Cloud and virtualization = disaster recovery. While not entirely accurate, the key idea is that virtualization (which makes production servers more portable) and cloud infrastructure (which provides an economical secondary location) can enable enterprises of all sizes to achieve rudimentary disaster recovery. This is especially true for midsize organizations that previously didn’t have a secondary venue to use for BC/DR, while enterprises often have other options.

Backing up SaaS

While most of this material presumed that the production servers were traditional on-premises resources, many of those workloads are starting to move to the cloud, including email platforms, CRM systems like Salesforce, and file sharing. Unfortunately, many SaaS products have not yet developed the APIs to enable traditional third-party backup developers to extend their enterprise backup coverage for the SaaS platforms. Historically, these APIs come as the platforms grow in mainstream use — but seldom soon enough. Without those APIs, traditional backup developers have typically been slow to add those SaaS offerings to their coverage areas. Because of that, it is not uncommon for new backup products to come from startup companies. For example, when VMware hypervisors were gaining initial popularity, it wasn’t the traditional physical server vendors that first mastered VM backups — instead, Veeam, PHD and Quest brought the first products to market. Later, when the APIs were released by VMware, the legacy products raced to embrace the capability and catch up to the early disruptors. It is likely that that pattern will repeat itself as early innovators are delivering new approaches for protecting SaaS products like Salesforce (CRM), Office365/Google Docs (file) and email services.

Any way that you look at it, the cloud will likely be part of every data protection and disaster recovery strategy, but whether backing up to the cloud or from the cloud, the approaches will vary dramatically.

 [Originally posted on TechTarget as a recurring columnist]

Leave a Reply