Ikea Up in a Cloud (of Smoke) – Is BC/DR Going to the Cloud?
By Zerto, on 22 February, 2010
A few weeks ago, a local Ikea megastore went up in flames, literally. Luckily nobody was hurt, but the huge building was destroyed, with all the furniture evaporating in a cloud of smoke.
One interesting anecdote, from a BC/DR perspective, was a small paragraph in the local paper describing how the IT team rushed in with the firefighters to salvage some of their hardware from the flames. The reason quoted by the reporter was their need to recover payroll and HR data…
The week of the Ikea fire, I joined ZeRTO’s product team for a road trip to meet architects and CTOs of the top cloud providers (two of them have merged since…). Different providers are in different stages of rolling out their cloud offering, some still building V1 and some already working on the second version. However, they all share key pieces of a common cloud vision.
One common denominator is their view regarding the future of BC/DR, and how the cloud will play an integral part in that future.
Thinking about it, it really makes sense. Imagine a Director of Infrastructure in an enterprise, being handed the task to provide or expand their DR. This requirement often comes from the CEO or the board, so it receives the top priority. The textbook way of doing it involves getting a DR site, either owned or a co-lo, purchasing all the necessary hardware and software (CapEx…), installing it, configuring it, managing it, buying the communication pipes, configuring everything… Very expensive, very complex and an excellent trigger for considering alternative solutions!
But in the ‘new age’, aka cloud, there will be easier and simpler alternatives:
Option 1: Moving the complete infrastructure to the cloud and let the cloud provider worry about BC/DR – this is the Holy Grail for cloud providers. However, this is not a realistic option for most enterprises today, as they do not yet feel the cloud is adequate for their enterprise-class needs. Moving production to the cloud will take time, just like virtualizing mission-critical production applications is only now becoming a viable option.
Option 2: DR to the cloud (or DRaaS – DR as a Service) – point your replication solution to a cloud provider, sign an SLA, and that’s it!
Well, DR to the cloud could be a true win-win. For the enterprise, it’s a simple way to achieve high levels of resiliency, but without all the complex and expensive processes required to build and manage their own DR site. For the cloud provider, it is a great first step to get enterprises to start moving critical applications into the cloud, avoiding several reservations associated with moving production itself to the cloud.
But there are some challenges. If we consider enterprise-class DR solutions, of the kind that usually happens at the SAN level (consistency requirements, low RTO/RPO), these solutions require both ends of the DR solution to have similar hardware equipment, which is not really an option for a cloud provider. Some cloud providers do offer this solution today, but due to its nature, it is mostly offered as one-off professional services engagements ($).
Another set of challenges arises from multi-tenancy and manageability from the cloud providers’ point of view. Can they, or should they, over-allocate recovery resources? What happens if there is a regional disaster and several of their customers have to failover at once? How do they protect, and more importantly, recover data from many customers efficiently while maintaining the security levels and SLAs they have promised?
And there are additional challenges. To be cost effective and competitive a cloud provider must be able to set up the service quickly, and without re-architecting the customer’s environment. Additionally, enterprises expect all cloud services to be consumed by subscription. Can cloud providers deliver this with the cost structure of their current infrastructure solutions?
The way I see it, DR to the cloud is going to happen, big time. I’m not the only one seeing this, judging by what I read from Gartner, Forrester, IDC and others. I spent a few days at VMWare Partners Exchange last week (more on that in a future post) and everybody I discussed this vision with agreed. How quickly the vision will be realized depends on the time it will take for the cloud providers and technology vendors to overcome these challenges.
BC/DR is changing, the cloud is changing, and we will see it taking off in the next few years!
On a side note, I have been writing recently about virtualization, how it changed the world, and claiming that storage will have to change accordingly. Check out what Steve Duplessie posted in his blog – similar content, but in his fun and colorful style.