Data defeats the cloud cloak of invisibility

Invisibility. A word reminiscent of magic and fairy tales is one of the latest terms that attempt to replace the overused and amorphous term “hybrid cloud.”

I’ve used a term similar to invisibility, but the whole notion is how to make the cloud invisible with respect to the corporate datacentre – not a separate infrastructure hundreds or thousands of miles away. Regardless of the moniker given, the datacentre and the cloud should seem like one contiguous entity.

There are three major challenges when you want to give the cloud a cloak of invisibility: networking, portability, and data. While many of the issues have been tackled, one is continuing to cause a wrinkle, or better yet a tear in the “cloak.”

Networking: Be transparent

The first challenge being addressed is networking. AWS has had Direct Connect for quite some time to provide a dedicated network connection between a corporate datacentre and AWS cloud datacentres. Now we are seeing some major industry vendors also attack this problem. VMware essentially abandoned its own cloud business, vCloud Air, and now is focusing its cloud efforts on using NSX. NSX, its Software Defined Networking Solution (SDN) offering, will now connect corporate datacentres to the cloud.

Microsoft is still going strong with Azure, but has introduced the Azure Stack to provide connectivity between Azure and corporate datacentres. Even Citrix has jumped into the mix with its CloudBridge solution. With other SDN vendors following suit as well as a number of VPN and private line solutions out there, we’ll assume that the networking challenge in the quest to make the cloud invisible is either solved or well on the way to be fixed.

Portability: Run your apps anywhere

The second challenge being addressed is portability. In the launch of Oracle cloud services, Larry Ellison discussed moving an application from on-premises into the cloud and from the cloud back to on-premises. He said, “You can move both directions.” Oracle also recently acquired Ravello, which enables existing on-premise VMs to run in AWS and Google.

VMware is also focusing on this area and made a joint announcement with IBM to allow customers to extend their existing workloads from their on-premises datacentre to IBM's SoftLayer cloud. Nutanix is now promoting an “invisible infrastructure” with a software-defined box that other hyper-converged vendors will likely emulate. The goal is to enable workloads to move between hyper-converged “boxes” whether they are on-premises hardware or cloud software.

Now workloads can run where most appropriate, the datacentre, the cloud, or both making the cloud invisible from an application perspective.

Data: Keep it close by

The gravity of data is what breaks this model, essentially defeating the cloak of invisibility with the cloud. This is because co-locating data with only applications in the datacentre is great for users of those apps, but users of cloud apps suffer with performance. Co-locating data only near applications in the cloud is great for users of those applications, but users of applications in the datacentre suffer with performance. Users and applications need to have hot data next to them at all times, whether in the datacentre(s) or in the cloud.

One way of solving this dilemma is copying or replicating data between the cloud and the datacentres. This would ensure that an application or user does not have to traverse a great distance in terms of latency to reach the data. In other words, an application in the cloud does not have to cross a high-latency, low-speed Internet connection to get to the data in a corporate datacentre and vice-versa.

While copying or replicating the data solves the performance issue with latent networks, it introduces file integrity issues as different versions of files are created. This may sound like just a nuisance, but it becomes a real issue when a building is constructed without the fire protection system because the wrong version of the plans is used. The same sort of issues occurs when changes to a software build are lost or when changes to an important document are overwritten.

This is not that different than what happened many years ago when each user had direct attached storage. Different users could be working on the same file at the same time in their own storage area creating different versions of the file that would later need to be reconciled. The shift from on-premises direct attached storage to on-premises shared storage back then is not that different than what we are discussing here. Instead of having separate islands of storage, there needs to be common storage than spans across the datacenter and the cloud.

Then, as now, there still needs to be a mechanism even with shared storage to prevent file corruption. The solution to the problem then was simple – lock the file when opened so that only one user or application can make changes at a time. We just need to extend this same capability now for storage that is across the cloud and the datacentre.

Spanning data across the cloud and the datacentre and enabling file locking between the cloud and the datacentre eliminates the effect of gravity on data. Data can now be hot in both the datacentre and the cloud providing apps and users in both locations with the same local performance without data integrity issues.

With the addition of data spanning across the cloud and datacentre to networking and automation/management, the cloud can truly be invisible – one seamless environment where it is hard to determine where the datacentre stops and the cloud begins.

Barry Philips, Chief Marketing Officer, Panzura

Photo Credit: T.L. Furrer / Shutterstock