While virtualization technology is very helpful, the indiscriminant use of virtual machines (VMs) can create new problems for a data center. The issues these organizations eventually face are how to manage a growing list of virtual servers and how to rapidly and automatically deal with the growing network, storage access and management issues created by the proliferation of virtual servers.
Flexibility is the goal of server virtualization
It's always important for business and IT decision makers to not just effectively deploy virtual machine software, but to keep revenues and profitability in mind. Time to market and time to profit are increasingly becoming major drivers in technology adoption.
With this in mind, the adoption of virtual machine software or any other type of technology is a means to an end, but not a goal in itself. What is the real goal, though? The end being sought is increased business flexibility and better service to the customer. Virtual machine software is being deployed because it can help an organization quickly set up and tear down the systems that support its rapidly changing needs in the most efficient way.
It is clear that virtual machine software which enables server virtualization was a great start at achieving the goals of organizational flexibility, efficiency and agility. That said, no one starts down the path toward a virtualized environment by saying, "I want virtualization." Users usually look for ways to rapidly deploy workloads, the storage needed for the applications and data making up those workloads and, of course, making sure that individuals can access those applications quickly on whatever system hardware the organization owns.
The limitations of server virtualization
While server virtualization certainly helps organizations achieve the goals of adaptability and mobility, virtual machine software is sharply limited. Before VMs can be initially deployed and later moved, a number of steps need to have already have been accomplished, including the following:
- installing the underlying physical machines with a hypervisor
- provisioning each of the VMs with the appropriate operating system, data management, application framework and application software
- granting access to all of the virtual machines to the appropriate storage facilities
- granting access to all of the virtual machines to the appropriate network facilities
Accomplishing these tasks can be a logistical headache and security nightmare. Without the help of automated tools, these simple steps would become time consuming, prone to error and would inhibit the organization's progress towards a flexible, adaptive data center. That is why most organizations pre-allocate VMs and leave them in place and running. They face significant problems when there is a system, storage or network outage--planned or unplanned.
I would recommend that organizations understand the ramifications of virtual machine software to create a server virtualization environment. If an organization hopes to deploy a large number of virtual servers to replace some of its physical servers for consolidation and agility, it's important to also implement server repurposing technology to address the basic physical bare-metal, network and storage access issues.
Repurposing is really a form of re-deployment. This can be a very difficult task to accomplish quickly and efficiently. Data centers for medium-sized and large organizations are usually constructecd as a number of separate subnetworks to reduce management issues. Moving systems and storage around in such an environment can be a daunting task.
Once a virtual machine software product such as VMware's ESX Server is deployed, organizations find that some applications or workloads must be hosted on physical, not virtual, systems. Some workloads or applications simply do not fit within the constraints and performance profile of virtual systems. Moving these workloads or applications adds to the complexity of system repurposing.
To use server virtualization in the most effective way, it must be possible to rapidly change which servers are running, what software stacks those servers are running and how those servers are connected to network and storage. It must be possible to make changes without also being forced to take up time consuming and error-prone manual tasks such as reconfiguring physical machines, cable infrastructure, LAN connections or SAN access.
I suggest that decision makers would be well advised to include server repurposing and its requirements in their planning regardless of whether the organization is planning to deploy virtualization technology from Microsoft, VMware or XenSource. Several alternatives are available though, including Scalent Systems Virtual Operating Environment (V/OE) software, Egenera's hardware based offering and Unisys uAdapt software.
About the author: Dan Kusnetzky is Principal Analyst at the Kusnetzky Group. Prior to that, he was executive vice president of corporate and marketing strategy for Open-Xchange and an IDC analyst covering operating environments and virtualization software.