A funny thing happened on the way to virtualization: Many IT managers discovered that instead of bringing simplicity to their busy data centers, the technology actually made things more complicated. Sure, physical server counts were reduced, energy costs lowered and hardware footprints shrank-all thanks to virtualization.
However, complexity and confusion grew, and managing virtual environments became an almost unbearable challenge. The situation has left many with false assumptions, technology myths and, worst of all, a sense that virtualization can cost more than it’s worth.
After all, the goal for most IT shops has been to do more with less, not less with more, and virtualization platforms have come to represent increased management needs, as well as higher operational overhead. Even so, the adoption of virtualization solutions grows, as does the dissatisfaction with how much savings can be delivered.
The truth is that this is not a technological problem, but a management problem. Luckily, that problem can be solved with technology, specifically virtualization management technology. However, before diving into that that aspect of virtual environments, IT professionals need to learn what the facts and myths of virtualization are before taking a deep dive.
One of the first things an IT manager should learn is that deploying a virtualization solution is never a quick fix for other problems.
Frank Marshall, director of global retail support for The Est??«e Lauder Companies, has seen firsthand that deploying virtualization solutions is not without its challenges.
“Virtualization can present as many challenges as it solves,” said Marshall. “Management becomes more complex, thanks to hypervisor issues, platform changes and all of the associated components needed to build a virtualization platform. It takes a holistic approach, where virtual machines are monitored, managed and maintained to truly benefit from everything virtualization has to offer.”
Deploying a Virtualization Solution Is Never a Quick Fix for Other Problems
During a recent presentation, Gartner analyst Thomas Bittman stated, “There will be more VMs [virtual machines] deployed during 2011 than in 2001 to 2009, combined.” Bittman warned that when deploying virtualization, an approach that is “too cold” won’t achieve potential savings, while an approach that is “too hot” will lead to virtual server sprawl, along with cost overruns and breakdowns.
Simply put, virtual server sprawl will become one of the biggest management issues for modern IT departments. Bittman added that management is the key to maximizing any virtualization deployment within an enterprise data center.
Srinivas Ramanathan, CEO of eG Innovations, which develops virtualization management software, agreed. He said, “You can quickly negate any savings if you don’t measure and manage virtual environments properly.”
The main point of all this is that IT administrators have to show where the savings are. The department must be able to measure those savings, and that requires a management approach that incorporates the ability to monitor and effect change at every level of a virtualization deployment.
That means IT departments must be able to quantify loads, capacities, demands and usage, as well as track trends for those elements. The problem, according to Ramanathan, is that most of the management tools included with virtualization products lack critical capabilities, meaning that those seeking holistic management must turn to third-party solutions.
“Management is even trickier in some cases,” said Ed Laczynski, vice president of cloud strategy and architecture at managed services provider Datapipe. “You are introducing a completely new layer. This is where we think virtualization as a service and infrastructure as a service works for enterprises that don’t want to focus on undifferentiated technical chores.”
Effective management is critical for IT departments that have to show a decent return on the company’s investment in the technology.
IT Administrators Have to Show Where the Savings Are
Gartner finds that effective use of virtualization can reduce server energy consumption by as much as 82 percent and floor space by as much as 86 percent.
Laczynski agreed with Gartner’s numbers. However, he added, “Without effective management of cloud and virtualization technology, ROI is difficult to realize and even harder to measure.”
For his part, Ramanathan claims that Gartner’s numbers are only the tip of the iceberg. He believes that an effectively managed virtualization deployment offers additional savings in the form of administrative overhead since management tools enable rapid deployment and repeatability, enhance conflict resolution, improve use of resources and speed consolidation.
Those elements, combined with power savings and footprint reductions, can equal big savings.
However, virtualization has a dark side.
One myth, according to Ramanathan, is that that virtualization makes management easier-a perception fueled by the fact that there are fewer physical servers to monitor.
In actuality, management becomes more complicated since virtualization reduces the physical footprint but increases the number of operating systems and applications that have to be managed. What’s more, a lot of changes in a virtualized environment, and the tools designed to help monitor and manage physical infrastructures, are ill-equipped to deal with the needs of virtualized infrastructures.
Virtualization Actually Makes Management More Complicated
Another myth involves the belief that that resource reservation simplifies virtualization management. In other words, it is thought that specific resources can be assigned to specific virtual machines to prevent conflicts and interference.
“Resource reservation may reduce risk but requires more specific knowledge of logical and physical infrastructure, introducing more complexity,” said Laczynski.
In reality, resource reservation is not always possible and it is far from efficient. While CPU and memory can have static assignments for a virtual machine, those resources are better used for dynamic loads to reduce waste. What’s more, other elements, such as disk and network, cannot be statically assigned to a virtual machine, further complicating the management conundrum.
A third myth is that virtualization platform metrics are sufficient for effective management.
However, native management platforms only focus on a single dimension, specifically the number of physical resources a virtualization platform uses. In reality, according to Ramanathan, multiple dimensions need to be managed and monitored, such as how much of a physical CPU is being used by an individual virtual machine and how much of those resources are being used by the applications being run inside the virtual machine.
The only way to get the full management scope of a virtualization solution is to know what is happening on the virtualization platform, as well as in each and every virtual machine running on the platform and every application being run on those virtual machines, said Ramanathan. That information proves critical for assigning resources, scaling systems, managing hardware use, as well as maximizing the savings that can be offered by virtualization.
Another false notion is that virtualization is another IT silo. Simply put, this means that virtualization falls into the hands of IT and is then segregated into its own management group, where the virtualization team manages the VMs and operates independently of the enterprise operation team.
While some think an independent VM management team helps expedite the rollout of a VM platform, the truth is that virtualization is a critical component of the overall IT infrastructure, and a problem with the virtual infrastructure affects business services. That means virtualization management should be integrated with overall IT management, allowing a unified approach. One way to manage all this is to invest in cloud infrastructure that someone else, like a Datapipe, runs off premise.
“However, managed service providers can provide the benefits offered by both physical and virtual servers, without introducing sprawl, by incorporating the most appropriate solution to a customer use case,” said Laczynski. “There is a better way to operate, where businesses aren’t worried about sprawl at all, or the complexity of it, because they use providers to manage that problem space for them.”