Well, here we are nearly in 2012 and a few years into the cloud storage movement. It’s amazing to look and see how many players have entered, and still plan to enter the market. I have a Google alert for “cloud storage” and it seems that every other week another company is announcing their new cloud storage service.
Now, to be fair, the definition of cloud storage has become quite muddy. If data is stored offsite, it’s now vogue to call it “a cloud” whereas in yesteryear it might have just been an Internet service, an application service provider, or software as a service. In my book Apple’s iCloud, Salesforce.com, and the seemingly dozens of “backup to the cloud” services are not really clouds at all. To qualify as a cloud storage service in my book you have to meet three criteria:
- Your service has to be open to anybody for any purpose. If “the cloud” is only there to support other offerings from the “cloud” provider, then it isn’t a cloud.
- Your service has to be reservationless and pay-as-you-go. Buying a managed LUN from your hosting company doesn’t count.
- Your service has to support at least two replicas of data in two different data centers. If you’re not capable of surviving a data center disaster without losing customer data, you don’t count.
Who offers cloud service like this? Amazon S3, AT&T Synaptic Storage as a Service, and Nirvanix, to name a few leading providers. These companies provide what is analogous to a never-full disk drive in the sky, that won’t lose your data.
End-users are starting to realize the benefits of the cloud as compared to managing their capacity on-site. And a number of companies have emerged to help make connecting to the cloud easy and secure, while providing a familiar and high performing user experience. These companies in effect take the disk drive in the sky created by the cloud service providers and turn it into a storage array with familiar provisioning and data protection features like snapshots, all accessible over standard storage interfaces such as iSCSI or CIFS, rather than having to program to the cloud providers’ native HTTP/REST APIs. This is good and it is an important step forward.
But it isn’t the end game. I’ve had the privilege to speak to literally hundreds of IT end-users at large corporations about the cloud. An interesting question recurringly pops up that I think speaks to people’s latent, as yet unfulfilled desires for the cloud. The question goes something like this, “Once I put my data in the cloud, can I access it from anywhere?” This makes perfect sense. If the data is in the cloud, then it should be reachable from any place with an Internet connection. True enough.
However, there are a couple of caveats. If the data got into the cloud through an on-premise gateway like I described earlier, then it must be read back through such a gateway as well. This is because the gateways encrypt data stored in the cloud and typically deduplicate it as well, so what’s stored in the cloud doesn’t resemble the original data anymore. It must be reconstituted through a gateway to be utilized. This is perfectly fine and within the realm of customer expectations.
The second caveat is the killer. Data ownership must reside with a single gateway at a time. This means that while a second gateway can access the active data set in the cloud in a read-only mode, it cannot write to it. This squashes the dreams of IT managers who envision having users at their geographically dispersed offices all over the world collaborating on data sets that are centrally stored, managed, and protected in the cloud. IT managers hate having to replicate and synchronize data, and they hate it even more when end-users go “off the grid” and ship CDs or disk drives full of data around. But since cloud gateway solutions don’t know how to manage file or record locking to prevent multiple-writer data corruption, they simply don’t allow multi-point read/write access to the cloud data.
However it doesn’t need to be this way. The technology exists to build multi-site file services at the network edge which leverage a common, centralized core – the cloud. Imagine how much easier it would be for IT to manage all their capacity in one place. Imagine what end-users could do if everybody at every site saw the same view of the file system, and could read and write to any data (given appropriate access controls, of course) no matter where it was created. This isn’t a disk in the sky. It isn’t a storage array in the sky. It’s a brand-new capability that didn’t exist in the past and can only happen by leveraging the power of cloud storage in new and unique ways, forgetting the old paradigms.
So to the end-users out there, keep asking your questions because they’re right on the money and vendors need to hear what you want. And to the storage vendor community, it’s time to think about cloud storage as an enabler for new and unique services, not just as disk drives in the sky.
