IT Infrastructure, Server Virtualization Management Developments Get Attention at Gartner Conference

While virtualization, especially in the context of virtualization within x86 hardware, remains one of the most important technologies being considered by IT managers right now, the challenges that virtualization will present in the next decade require strategic planning now.

At the Gartner Symposium/ITxpo 2008 in Orlando, Fla., the research firm’s analysts were detailing a vision of how IT managers will be able to create virtual environments in data centers that have a mix of hardware and operating systems. While most of virtualization has focused on how to create virtual machines within x86 hardware–especially when one considers what VMware, Citrix and Microsoft have done in the past two years–there is a matter of how those virtual environments will work and integrate within data centers that also utilize mainframes and Unix-based servers that support legacy applications.

While trying to deal with virtualization in these types of heterogeneous data centers, IT managers are going to be confronted with a number of choices and scenarios for how to best maintain and control virtual machines and applications. The question that IT will have to confront will range from whether to stick with one virtualization vendor, such as Microsoft or VMware, or whether to place applications in the virtual environment that best fits that workload, whether it's Unix, a mainframe or x86 hardware.

“The issue here is, what am I faced with in terms of what do I want to achieve and what type of optimized capabilities do I want?” George Weiss, a distinguished analyst with Gartner, asked in his Oct. 14 talk. “For example, if I want to do optimized placement of my workload, it is a noble goal, but what platform should I use? So you really have to make decisions about whether it’s going to be on large vertically integrated SMP [symmetric multiprocessing] systems, or is it going to be on cluster scale-out types of technologies.”

While it’s possible to move virtual machines around and fit them within the right platform now, Weiss that in the future, a lot of these decisions are going to have to be automated with the virtual machines finding the right platform that will optimize the workload.

The problem with these and other goals is that management software for virtual environments is only being developed now and many still only address management of homogenous environments such as all virtual machines on x86 hardware. There is also a conflict of how a company’s data center will grow during the next 10 years. There is the scale-up model, which provides for more integration but is costly, or the scale-out model which is cheaper but the components are not integrated as well.

While there are many providers of analytics, lifecycle management and resource management software out there now, Weiss believes a few companies have started to offer the beginnings of what will eventually turn into management tools that will address virtual machines in heterogeneous data centers.

For example, Hewlett-Packard, with its Insight Dynamics VSE software, and Sun Microsystems, with its xVM Ops Center suite, are each offering ways to manage both the physical hardware and virtual environments. On the other hand, IBM is developing a management platform called “Ensembles.” While Big Blue executives have not talked much about this platform, it has the potential to segment the data center into common resource pools and allow IT to draw compute power from these pooled resources.

There are also a number of smaller vendors offering a range of different software that helps address some of the problems associated with managing a heterogeneous data center. One example is CiRBA, a company that provides analytics modeling.

While Weiss urged the audience to look at and evaluate these different vendors, he cautioned that the market is crowded right now and many of these smaller virtualization management companies will simply go out of business or become acquisition targets for the large vendors.