Get Those VMs Under Control - Systems Management News On The Web

Sort By :DateRelevance

DEPARTMENTS

HOME

TOP STORIES

ON THE WEB

BZ MEDIA

Printable version

Get Those VMs Under Control
By Michelle Savage

June 30, 2008 — IT departments are increasingly implementing virtualization as a way to lower the number of physical servers and ultimately reduce power and cooling costs. However, virtualization experts warn that if virtual machine (VM) management practices aren’t put in place from the start, data centers will suffer from VM sprawl.

Because virtualization makes it easy to deploy VMs, it’s easier than ever for servers to grow like dandelions—a phenomenon known as VM sprawl, in which VMs become unmanaged and uncontrolled as they pop up throughout the IT infrastructure.

Anthony Mar, product marketing manager at Embotics, a virtual server management company, said companies stand to lose thousands of dollars if they allow the number of VMs to grow too quickly without proper care and feeding.

Mar described VM sprawl as proliferation without adequate IT controls. He said that one of the biggest problems with virtualization is that many organizations mistakenly think that once they purchase and set up the basic virtualization infrastructure, rolling out additional VMs is free. However, applications still require processing, memory, storage and networking—all of which cost money.

Mar said these challenges are relatively new to IT managers as they did not exist in the physical environment. “Physical server costs in data centers have a built-in limit,” he said. “As they proliferate, eventually you simply run out of space or budget for power and cooling. If left unchecked, the costs due to virtual server sprawl will eat up your entire software license budget, more and more administrator time, and eventually require more physical servers. It's a situation that worsens as IT budgets grow.”

Keep Track of Your VMs
According to David Link, CEO of ScienceLogic, another major challenge of virtualized environments lies in knowing what’s where. He said that the rapid growth in VMs makes it necessary to plan, deploy and optimize the performance of virtual environments. “Even when you create high availability with VMware, you still need to know what’s going on in the hardware beneath the hypervisor, how the guest operating systems are behaving and how your applications are running. With multiple VMs, it’s harder to get a true representation of what’s connected to what.”

Without a good management tool, Link said IT managers are “driving in the dark without the lights on.”

In a physical environment, all of the systems, network devices and storage units are organized and fairly easy to find. But VMs are ephemeral and are generated, provisioned, produced and moved around constantly. This makes it hard to tell what applications are running, where they are located, who owns them and what they are doing.

When it comes to implementing a virtualized environment, such disorganization can be detrimental to an enterprise. “Some companies just slap it up there without the planning or thought process,” he said. “They don’t take the time for operational metrics and view of overall health. Their monitoring systems are the end users. Fortunately, we’re seeing less and less of this. Most seasoned IT professionals want to know how each guest operating system is doing, especially as users become less tolerant of applications being down. It’s a massive productivity loss.”

Without monitoring and management tools, companies will find it hard to manage virtual environments in a world where everyone expects things to work perfectly from an IT perspective, warned Link. “Without the details of how things are behaving, performing and changing, you get caught flatfooted to deliver the level of service that’s expected today.”

In addition, without tools to help discover, track and manage virtual resources, Link said that companies waste a lot of time and money trying to figure out what these tools could quickly and easily reveal.

“Some of our customers generate billion pieces of information about the health of their systems a day,” he said. “[ScienceLogic's] EM7 takes those pieces, correlates them and filters them into 100 things that need attention. Without monitoring tools, these companies would need to manually look at things to inspect the health. It would be impossible.”

As more VMs are added, the need for management increases. “When you get to scale, the problems get exponentially harder,” Link said. He was at a recent conference at which a U.S. government official described reducing a Pentagon data center from 130,000 square feet to 8,000 square feet through virtualization. “There are many good rationales to reduce your servers. But companies need to know that this doesn’t decrease the complexity—it increases monitoring complexity. You need the right tools to monitor.”

Gartner vice president and analyst Thomas Bittman said that good VM management is needed from the get-go. “Virtualization without good management is more dangerous than not using virtualization in the first place.” He advised companies to have a good management strategy and vision for their VMs, including plans for configuration management, change management and capacity planning.

Balance Virtual Traffic for Efficiency
Understanding peak utilization trends is another important consideration in virtual environments. It is even more critical in virtual environments than in physical ones because so many more servers are competing for resources, said Dave Asprey, vice president of marketing at Zeus Technology.

Particularly in the Web services industry, many companies find it difficult to monitor and load-balance massive increases in Web traffic across their networks, said Asprey.

“One of the biggest concerns with virtual environments is the fear that a company will lose its application if the hardware goes down,” he said. “That’s why companies need two pieces of hardware running VMware, and software that makes sure traffic gets from one server to another if one breaks or slows down.”

Naived Merchant, operations manager at Accenture, a technology consulting firm, believes that while the cost of running a virtual environment is much lower than that of a physical one, the savings can be eroded when the virtual environment is not managed properly. However, he stressed that management solutions are useless without a good team in place to implement and manage these virtual environments.

“Monitoring 24 hours a day, 365 day a year is key to properly managing a virtual environment, as there are so many factors that are important in that type of environment,” said Merchant. “Most companies use tools to ensure uptime and availability of critical applications, and this is very important. But we haven’t gotten to a place where applications and technology are self-healing.”

Merchant said that management tools are great when it comes to finding out what’s wrong early, but they are inadequate when it comes to getting that problem fixed. “Monitoring is a catalyst for change management teams,” he said. “If they see that something is going to happen, they are the change agents that bring everyone together and play role of collaborator. You need these teams to understand what will be impacted and pull in the right people from different teams to address the problem.”

The bottom line for IT managers is that multiple VMs can be controlled if companies take the time to map out their strategy and vision, invest in automated, policy-based management for their virtual worlds and have good teams in place to manage the complexity of these environments.

Related Search Term(s): Server management, virtualization

Share this link: http://www.sysmannews.com/link/32453

EMAIL THIS ARTICLE

SEND FEEDBACK