By now, it should be clear to just about everyone that the simplified management claims of the early virtualization systems were, to be generous, not entirely accurate.
Sure, provisioning and deployment became a snap, but once the virtual environments were up and running, it became difficult to map data to storage, monitor use, find and decommission idle VMs and a host of other crucial tasks. To their credit, each of the top virtualization vendors has sought to address these issues in subsequent releases, but the issue of virtual management is about to take another twist: the rise of heterogeneous virtual environments.
As a new survey from The Info Pro has shown, while VMware still holds the dominant position among current virtual deployments, a growing number of enterprises are opening up to the possibility of launching rival platforms, namely Hyper-V and XenServer, as well. The company reports that nearly two-thirds of VMware users have already tested alternate environments, with nearly a quarter planning to launch a non-VMware system in the coming year.
That means many organizations face the prospect of maintaining separate virtual environments -- what some would call "virtual silos" -- an ironic turn considering that one of virtualization's initial claims is that it would break down the physical silos that have crept up in many enterprises to date.
The alternative, of course, is to find a management system that can span heterogeneous virtual environments. And fortunately, a number of third-party contenders are stepping up to the plate.
One of them is VKernal, which recently extended support for its Capacity Analyzer to Hyper-V. While not a full-blown management platform (yet), VKernal does provide a key function in that it can drill down into virtual environments to indentify which resources are underused at any given moment and assign VMs accordingly. The idea is to prevent stacking too many VMs in one area, leading to data bottlenecks in physical systems. With both ESX and Hyper-V in the mix, it should be easier to deploy both systems without them tripping over each other.
Management of virtual server infrastructure is one thing, but if the virtual desktop movement starts to pick up in 2010 as expected, expect a tough time of it managing umpteen desktop images under various platforms. That's why Red Hat has open-sourced its Simple Protocol for Independent Computing Environment (SPICE) format for hosting virtual desktops. Built on technology acquired from Qumranet, SPICE acts as a remote rendering system, allowing desktop images to incorporate high-bandwidth applications like multimedia and VoIP. As an open source solution, SPICE should find it easier to be incorporated into leading VDI platforms, making interoperability in heterogeneous environments that much smoother.
Meanwhile, companies like Dell are taking pains to include the virtualization layer as they seek to build full data center platforms based on the idea of preventing vendor lock-in. One of the ways it is doing that is the Advanced Infrastructure Manager system that is said to provide an easily configurable platform for a range of virtual/physical combinations that can adapt to rapidly changing workloads. The package aims to provide a dynamic management regime that can alter system configurations in a matter of minutes and can be integrated easily into existing heterogeneous environments, rather than rely on a customized solution built for a specific mix of technologies.
Simplified management has been the goal of IT since the founding of the industry, so much so that many people considering it to be little more than tilting at windmills. But complex management systems tend to arise from the fact that most solutions are geared toward a specific platform. The more platforms you launch, the harder it becomes to manage everything as a single whole.
But virtualization, and by extension cloud computing, offer a unique opportunity to break that pattern by shifting the focus off hardware and onto software. And that makes it a lot easier to break down the barriers that have made it so hard to match management and oversight with the need to scale up performance and capacity.