To get in the cloud, the enterprise first needs to build a functioning virtual infrastructure. This is not technically true, of course, but the management and operational challenges of a non-virtual cloud are so extreme that they make it practically unfeasible.
But once the need for virtualization has been established, the next question is, what kind? Not all virtual environments are the same, and it could be possible that the very solution that IT is rolling out to improve resource utilization and lower costs in the data center may not be the best approach for the cloud. And with IT complexity bordering on ludicrous these days, how can the enterprise be sure that a cloud-optimized virtual solution in one area will prove equally beneficial to another?
According to tech consultant Salman Ul Haq, the fundamental issues with virtualization do not disappear simply because it now supports a cloud. Virtual environments are more prone to system- and application-level threats and the extra strain that high utilization places on physical resources means that multiple applications and users are now prone to less reliable single-points-of-failure. At the same time, the cloud provides an entirely new playing field for users to run amok; spinning up new virtual resources at will and leaving IT to pay the bill. Tools like log file encryption and end-point data redundancy checks can help, but they must be instituted on the virtual layer, or within the hypervisors themselves, to be effective in the cloud.
At this point, both VMware vSphere and Microsoft Hyper-V seem to have taken the lead in adopting cloud-ready features, but has either proven to be the cloud solution of choice yet? It depends on your philosophy, says virtualization consultant Paul Schnackenburg. VMware sees virtualization as a base on which to build not just cloud architectures but all manner of workload management services, while Microsoft approaches virtualization as an extension of the operating system. These differences are evident across a number of parameters, ranging from memory management and storage support to fault tolerance and backup technologies, and Schnackenburg provides an interesting read parsing through many of these details. One thing is certain, though. Hyper-V’s feature set has risen dramatically in the past year and it still hits a lower price point than VMware, which means the latter’s dominance in the virtual arena could finally face a real challenger.
But virtualization won’t produce a cloud environment on its own. The other piece of the puzzle is automation, and this is already shaping up to be the next area of contention between the leading virtual platform developers. With tools like VMware’s Virtual Machine Manager and Microsoft’s Azure Services for Windows Server, the overriding goal is to foster seamless integration between physical, virtual and internal/external cloud architectures–essentially weaning the enterprise away from the data center and into a perpetually scalable, distributed computer environment.
These are still the early days for the cloud, however, so the virtual-cloud integration layer is anybody’s game. This fact has drawn a number of third-party developers into the field, many looking to capitalize on the latest software-defined networking (SDN) developments. One of the newest is PLUMgrid, which recently launched the latest version of its PLUMgrid Platform that forges links between virtual environments across disparate infrastructures. Running on standard x86 hardware, the system creates what the company calls Virtual Domains that can be used to build, configure and secure multi-tenant virtual networks on the fly. Once a virtual instance has been created, it can be deployed across multiple architectures regardless of their physical location.
For a long while, it seemed that virtualization was an end to itself–increase resource utilization and foster greater data flexibility. But utilization rates are about as high as comfort levels allow, and now it turns out that the flexibility side of the technology is where the real action is.
For the enterprise, then, the guiding principles behind future virtual deployments will need to shift from today’s budget needs to tomorrow’s cloud-based data ecosystem.