Subscribe and listen to AW’s podcast!
Subscribe and listen to the Automation World Gets Your Questions Answered podcast!
Listen Here

Four Tips for Successful Physical-to-Virtual Migrations

Only a fraction of process control users have adopted virtualization. A major reason why is that users are often looking for savings in the wrong places.

James Cage, automation solution integrator, Avid Solutions Inc.
James Cage, automation solution integrator, Avid Solutions Inc.

In 2003, a young software company called VMWare gave users a way to move a running virtual machine (VM) from one physical machine to another with no disruptions and zero downtime. VMWare was quickly purchased by another tech company for $635 million. Today it’s worth over $40 billion. Process control users look at the benefits of virtualization with envy. So the question is: When should you decide to migrate from physical personal computers (PCs) to virtual PCs? Following are four tips to help you arrive at the correct answer for you.

Tip #1: Use the right numbers
Are process control users adopting virtualization? To find out, I asked Jay Pritchett, director of engineering at Altran North America. He estimated that only 20 percent of the jobs he sees today are virtual. Pritchett explained that users often look in the wrong place for savings. “Don't expect initial costs to be less,” he said. “Just because you have five machines instead of 50 doesn’t mean your hardware price will drop. The boxes that are left will be costly, and you will also need to buy software and learn how to use it.”

But there are real savings. “The lifecycle cost can be very attractive because the cost of maintaining the system and deploying new or replacement nodes plummets, particularly in terms of the time needed,” he said.

Also consider availability. While the number of physical PCs may go down, Pritchett noted that “the availability of a system can go up.” In a physical system, replacing a failed PC involves finding replacement hardware, restoring a (hopefully) current image, and moving the hardware into place. At best, it takes hours. It’s also risky because no two physical computers are exactly alike, and small differences can cause the new node to fail. In the virtual world, you would simply restart the VM on a different physical machine, or “host,” a process that takes a few minutes. The restart can often be automated, cutting the time required even further. It’s less risky because the virtualization software protects VMs from differences in physical hardware across hosts.

But if you think this and other virtualization features aren’t as easy to implement as they sound, you’re right. It starts with hardware selection and architecture.

Tip #2: Hardware still matters
Virtualization solves many compatibly issues, but hardware remains critical. Remember, virtual devices just route requests to available hardware in the real world. It’s amazingly fast and efficient, but without the right underlying resources, the process breaks down. Efficient workload distribution means less overall processing power and memory, but you will need better hard drives. Virtual PCs are just sets of very large computer files. Without fast storage, your system will slow to a crawl when VMs start up, shut down, or move from one host to another.

While many architectural issues are intuitive (for example, never place both VMs in a redundant server pair on the same physical host), many aren't (see my earlier blog post for an explanation). So, where do you get the information you need?

Tip #3: Ask about vendor support
Every control system vendor claims to support virtualization, but support quality varies widely. “Ask a lot of specific questions,” said Pritchett. “What hypervisor have you tested? What licenses are necessary? What kind of storage is required, and what are the tradeoffs for different storage options? What network architectures are documented? And above all, what have you actually tested?” Answers should come from product documentation, not just case studies and brochures. Be sure your vendor provides the same level of tech support for virtual systems, and unless you enjoy finger-pointing exercises, choose one that supports the hypervisor and physical PC in addition to their own software and hardware.

Tip #4: Start small; start smart
Don’t get discouraged. Support among control vendors is rising all the time. You can learn more about virtualization and gain many of the benefits without spending a lot of money. VMWare Player is a free version of the most popular virtualization "hypervisor" software. Use it to create new virtual machines, convert physical PCs to virtual machines, and to run VMs created by others. For a few hundred dollars, products from VMWare, Microsoft and others allow you to use many VMs simultaneously and manage backups, snapshots, and cloning.

When you're ready to apply what you've learned, migrate any offline development, testing, and training systems to virtual machines. (If you don’t have offline systems, talk to your controls vendor. You should never have to pay full price to create an offline system.) Then, consider migrating a small part of your process, or making your next on-process system addition virtual. It's OK to mix virtual and physical nodes in one system.

But whatever you do, get started! The benefits of virtual systems are real, and if the challenges are formidable, well, so what? That’s just life in process control.

James Cage is an automation solution integrator at Avid Solutions Inc. He has served clients in a variety of industries for more than 24 years as an engineer, consultant and product manager. Contact James at or follow him on twitter at @JamesDCage. For more information about Avid Solutions, visit Avid’s profile on the Industrial Automation Exchange.

Barriers to cyber-attacks
How to apply defense-in-depth cybersecurity strategy to protect your company. Learn from your peers in industry.
Download Playbook
Barriers to cyber-attacks
Test Your Machine Learning Smarts
Take Automation World's machine learning quiz to prove your knowledge!
Take Quiz
Test Your Machine Learning Smarts