Is the cloud undermining enterprise disaster recovery?

By Jeff Vance

One of the main weaknesses of cloud computing is that it must rely on the Internet, which brings to mind Frank Lloyd Wright’s famous quote about my hometown of Pittsburgh. When asked how to improve the city, he quipped: “Abandon it.”

Pittsburgh’s infrastructure is a tangle of bridges, tunnels, inclines and steep roads that can be impassible with the lightest coating of rain or snow. In other words, for transportation, it is a mess. Similarly, the Internet was not designed to handle the volume and types of traffic now traversing it every second of every day.

For consumers, this is usually nothing more than a frustrating inconvenience. For businesses, though, the cloud combined with the fragility of the Internet could have disastrous ramifications. Let’s look at one of those disasters looming on the horizon. Many organizations are trying to shift disaster recovery (DR) to the cloud, and many cloud providers make it seem like DR is almost a checkmark feature. Unfortunately, it is not.

Recent outages at Google and Amazon, as well as recent natural disasters like the Fukushima earthquake and tsunami, point to one big flaw with moving DR into the cloud: regional outages and disasters can easily torpedo enterprise DR. For true DR, enterprises must spread their data across multiple data centers in different regions. The trouble is that this is not that easy to do.

Many services billed as DR are really just data replication services, meaning that end users aren’t able to mirror their infrastructures. Yes, users can replicate data, but not whole systems. OSes, apps and user preferences are not replicated. If a disaster happens, systems must be rebuilt before data can be recovered. This means you’ll experience even more downtime, which may well translate into lost customers.

Of course, it makes sense to focus solely on data replications. Sending that much data over the public Internet isn’t practical, while sending that much information over private networks, such as MPLS, is cost prohibitive – which is why so many businesses still handle DR physically, with tape backups.

The funny thing is that the problem of DR in the cloud runs parallel to another problem that hits distributed organizations: providing access to mission-critical applications to remote workers in branch offices. Branch office workers trying to access, say, an ERP application remotely will often experience latency that kills productivity. Deep-pocketed organizations dodge this problem by either dropping another app server in the branch, or they rely on expensive private networks. Either way, these companies pay a steep price to keep their branch office workers productive.

Another alternative is WAN Optimization. The trouble with traditional WAN Optimization is two-fold. First, most vendors require hefty up-front purchases, as well as requiring you to drop hardware into each site. What’s the point of getting rid of one box (an app server) for another? Yes, your application will stay in synch, but that seems like a steep investment for synchronization.

The other problem is that WAN Optimization is a highly complex solution. If something goes wrong, will the branch office have anyone who can fix the problem?

Fortunately, where the cloud presents a problem, it also offers a solution. A new class of WAN Optimization, pioneered by the sponsor of this blog, Aryaka, moves the complicated technology out of a dedicated box and into the cloud.

The benefits of this approach are many. First, costs are kept low because of cloud efficiencies. Second, users can consume the service based on their needs, not on nebulous projections. Third, because this is a cloud-based, highly efficient architecture, the overall solution can combine the benefits traditionally associated with Content Delivery Networks (CDNs) with WAN Optimization.

For instance, Aryaka’s global network is strategically built to ensure less than 20-millisecond latency to over 90 percent of the world’s business users. Redundant POPs allow for high-availability and seamless fail-over. Each POP location is connected to other POPs over multiple links and providers for increased redundancy and failover capabilities, ensuring five-nines reliability for the network.

Sounds like a description of a CDN, right? It’s not. That’s the backbone of Aryaka’s cloud-based WAN Optimization solution.

Combining three technological advances – cloud-based computing, WAN Optimization and a cloud-based private network – means that you can achieve true DR capabilities even for applications residing in third-party clouds.

Without an affordable, geographically dispersed network, it’s nearly impossible to achieve true DR in the cloud. With one in place, though, cloud-based DR can be consumed like other cloud solutions: as a cost-effective, efficient service.

Guest blogger Jeff Vance is the founder of Sandstorm Media, a copywriting and content marketing firm. He writes about cloud computing for a number of publications, including Forbes.com, CIO, Network World, Datamation and many others.