Manage Learn to apply best practices and optimize your operations.

Managing server virtualization complexity

Server virtualization, whether it's operating system virtualization, grid clustering or logical partitioning, adds complexity to any data center. Managing those environments can be achieved with emerging tools.

The variety of available server virtualization technologies provides both choice and flexibility for enterprises looking to optimize server utilization and performance. But as the deployment of server virtualization in production environments increases, the challenge of managing disparate virtual technologies are not virtual but are very real.

More on server virtualization in the data center:
Server virtualization: planning and pitfalls

Server virtualization myths and realities exposed

Virtual servers, like physical servers, still need to be configured. Patches need to be applied and performance must be managed. So although you may have fewer physical servers to manage as a result of server consolidation via virtualization, you still have to manage both the physical and virtual servers.

Virtualization: One type does not fit all

Before we continue, it may be helpful to take a brief look at some of the more prevalent types of server virtualization. This is not intended to be a comprehensive list of virtualization types, rather an introduction to provide some context.

First, there is system virtualization, which uses logical partitions with each logical partition running an operating system. For clarity, the term "host" to refers to the physical server. Furthermore, there is no underlying host operating system. The system manufacturer uses a native hypervisor to control the logical partitions. This type of virtualization is offered by server hardware vendors like Dell, HP, and IBM.

The next type of virtualization is operating system virtualization. The host server has its own operating system, and the virtualization software solution then allows multiple guest operating systems to run on the system. VMware and Microsoft Virtual Server are examples of this type of virtualization.

A type of virtualization that is unique to open source is para-virtualization, where the guest operating systems are modified so they can work cooperatively with the hypervisor. Xen is an example of this type of virtualization.

Server clustering is another type of virtualization. Clustering involves locally attaching multiple physical servers together, and they appear as a single processing resource. Server clustering is typically offered by server hardware vendors like Sun, HP and Dell.

There are also other types of virtualization, such as grid computing, but for the sake of brevity I'll limit the discussion here to the few types of virtualization technologies listed above.

Virtualization adds levels of complexity

There's no question that virtualization offers valuable benefits for IT organizations but it also introduces additional levels of complexity. First, consider the different types of available virtualization technologies spanning from server clustering, to logical partitions, to operating system virtualization and more. Next, add the number of available virtualization solutions supplied by a variety of vendors. Then, figure in the abstraction layers, as well as the dependencies of the virtual and physical entities. After that, consider the potentially dynamic nature of virtual servers as work loads are shifted from server to server for performance reasons. This is in addition to physical server additions and replacements. Finally, all of the servers, both virtual and physical must be tracked, managed, maintained and their performance managed. When considering all of these factors, the complexities become apparent.

As enterprises expand their use of virtualization technologies in production data center environments, the reality of managing large-scale, heterogeneous virtualization technologies present a significant challenge. Although server virtualization solves critical issues for IT, it also introduces management complexity because of the different approaches and vendor platforms involved.

Different server virtualization approaches are very distinct and have unique characteristics and management requirements. But there are virtualized server solutions for different server platforms, such as Red Hat Linux, Microsoft Windows, IBM, SunFire Midrange and High-end Servers, Veritas Cluster, and HP. However, as enterprises deploy more virtualization solutions that are scattered across the enterprise, managing the disparate virtualization technologies using fragmented element-specific tools becomes a challenge in itself.

New virtualization management products emerging

Ptak, Noel and Associates (PNA) sees the general state of the virtualization management segment as still immature and developing. There are fragmented solutions that are still evolving. In the past, many management vendors simply added support for virtual servers by adding discovery and some management capabilities to their existing management products. PNA is now seeing a new breed of virtualization management solutions emerging as vendors are starting to deliver better, consolidated solutions that are specifically designed to manage the unique characteristics of the different types of virtual servers and the underlying physical servers that they depend upon.

Vendors such as CA, HP, BMC and Opsware now have management solutions specifically geared toward virtualization environments. These new solutions begin to address management challenges that are specific to virtualized servers, such as visualizing and leveraging the relationships between the physical and virtual.

From an operational perspective, some of the solutions enable administrators to use a common set of commands or actions to manage the physical and virtual servers, collective displays and have consistent management capabilities across all server virtualization types. This reduces "management overhead" and costs. If administrators don't have to learn technology specific commands and interfaces to manage the different virtualization solutions and platforms, the ramp up time for new administrators shrinks and the time saved for experienced administrators can add up.

Another aspect of the complexity issue from the human perspective is that systems administrators who deal with virtualization must either become jacks of all trades or collaborate closely with other administrator experts to effectively manage the multiple layers of virtualization technologies. In a typical day you might be interacting with an astounding number of technology permutations:

  • wide variety of hardware platforms, e.g. Sun, IBM, HP, Dell, etc.
  • wide variety of operating system combinations, e.g. Windows, Red Hat Linux, Solaris, HP-UX, AIX, SUSE Linux, etc.
  • variety of virtualization solutions, e.g. VMware, Xen, Microsoft, Virtual Iron, etc.
Add to this multiple applications that interact with each of these components. If you're using multiple management tools, there's more to deal with. The bottom line is it's impossible for an administrator to manage this environment by him or herself. Becoming an expert in all of these areas is impossible. It must be a collaborative effort involving administrators doing some level of cross-training and tools to help lessen the learning curve.

Be sure that you have a virtualization strategy. An integral part of that strategy should be a virtualization management approach to ensure that you can keep it all under control and performing optimally.

ABOUT THE AUTHOR: Audrey Rasmussen has over 28 years of IT experience. She served as vice president at Enterprise Management Associates, a systems engineer at IBM, and co-authored the Network World Fusion Network and Systems Management newsletter for several years. Audrey is currently an analyst at Ptak, Noel & Associate.

Dig Deeper on Virtualization and private cloud

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.