Virtualization: Without Control, Power Is Nothing


Alan Maddison

June 2010

 

Summary: As a technology, virtualization provides some compelling capabilities. After it has been deployed, however, it will provide some significant management challenges. Implementation of the Microsoft Operations Framework (MOF) can help organizations overcome these challenges.

 

Introduction

As a technology, virtualization provides a fundamental shift in the way in which an organization manages its infrastructure. While this is true for all organizations, there is often a distinct difference in the significance of this change for a company, depending on how mature a company is in its use of virtualization. For organizations that are beginning to mature in their use of and reliance on virtualization, there is much more significance in this change in management requirements. This is primarily due to the fact that, as the use of virtualization matures within an organization, there is often a desire to apply virtualization to a much broader range of systems—moving from what is commonly referred to as Tier 3 or noncritical systems to Tier 2 and Tier 1 or mission-critical systems.

As the last barriers to virtualization are eliminated for the most critical applications, IT organizations are finally able to fully embrace and deliver on the concept of IT as a Service. Embodied most recently in the concept of the dynamic data center, this aspect of IT as a Service fully utilizes the entire suite of capabilities that are native to virtualization, including dynamic workload management, automated provisioning workflows, and self-service capabilities. However, with the achievement of these capabilities and the inherent fluidity and dynamism that by definition become a daily part of your IT infrastructure, there is a corresponding increase in the operational complexity of your environment.

As this change occurs, the need to have a clearly defined management framework that is capable of providing structure is imperative. Unfortunately, while many IT administrators understand that, once virtualization is part of the IT infrastructure, there is a critical need to ensure that the tools that are used to manage that infrastructure are effective, they overlook the importance of an effective management strategy. Implementation of a comprehensive management framework can not only facilitate operational goals, but it can also make a tremendous difference in ensuring that the organization achieves its long-term strategic goals. Perhaps more importantly, a failure to implement an effective management strategy alongside the right management tools can lead to virtualization becoming another difficult technology to manage—one that does not meet expectations and fails to deliver the expected return on investment (ROI).

This article takes a look at the challenges of virtualization management and understanding how the Microsoft Operations Framework (MOF), if applied correctly, can ensure that you are effective in addressing your day-to-day operational concerns.

 

Microsoft Operations Framework

Currently, at version 4.0, the goal of the MOF is to provide practical and relevant guidance on the delivery of IT as a service. By combining Microsoft best practices with a service-management framework and then mapping these guidelines to real-world processes, the MOF can be used to help align IT with the needs of a business. Because of the granular nature of the framework, it can be successfully applied either across the entire enterprise or to a specific service. However, regardless of your approach, the detailed steps that the MOF provides make the process of understanding your environment very straightforward, in terms of the MOF components. Moreover, the framework covers the entire life cycle of an IT service—from conception and development to operation, maintenance, and retirement.

The MOF is structured to support the concept that the life cycle of a service is based on three ongoing phases—Plan, Deliver, and Operate—with one underlying layer that operates over the entire life of a service—Manage. Each phase defines a number of goals that are achieved when the phase has been completed. For example, some of the goals of the Plan phase are to ensure that IT services provide value, are predictable and reliable, are cost-effective, and can adapt to changing business needs. To achieve these goals, each phase contains what are known as Service Management Functions (SMF), which define the processes and people that are necessary to complete each phase successfully and move to the next phase.

For example, the Plan phase provides a number of SMF, including Business/IT Alignment, Reliability, Policy, and Financial Management. Within each of these functions is a series of deliverables, each of which provides a clearly defined outcome. For example, the Business/IT Alignment SMF defines the deliverable of this function as being an IT service strategy. This delivery, in turn, will: allow IT services to be aligned with business processes and functions, define services that will directly support business needs, provide knowledge about how services will be used, and, perhaps more importantly, ensure that business units are satisfied. Oversight of all of these processes is provided within each phase through the use of Management Reviews.

Each phase includes at least one Management Review, to ensure that objectives are being met. By defining a series of controls, the MOF provides a mechanism that allows Management to provide guidance and ensure that each phase is completed correctly. Underlying all of these phases is the Management Layer, which serves  to ensure that best practices are followed, risk is managed, and  the IT service delivers the expected value. In addition to the value that the framework itself offers, there is a rich set of resources that complement and strengthen the MOF, particularly from a practical day-to-day operations perspective. Two of the most compelling sets of resources are the Solution Accelerators and Job Aids, both of which offer practical aids and guidance to real-world operational challenges.

 

Using the MOF to Overcome Management Challenges in Virtualization

There are many aspects of virtualization that make it compelling to IT organizations. Some of these reasons include reduced capital costs, improved operational flexibility, and efficient utilization of hardware resources. In fact, for most organizations, the initial impetus for virtualization often comes from this desire to consolidate servers and maximize the efficient use of hardware resources. As many IT  administrators know, with a traditional physical infrastructure, there is a strong desire—and, sometimes, a  technical need—to isolate applications and services onto their own hardware, to help avoid conflicts and maximize availability. From a business-operations perspective, many organizations see a need to isolate applications and services, because of the asset-ownership model that they use. However, regardless of these reasons, the end result is that it is very common to find that physical servers are tremendously underutilized.

Given the need of IT organizations to deliver services cost-effectively, this underutilization is always a tremendous concern. The attractiveness of virtualization and its ability to share workloads efficiently across the same set of hardware—while maintaining the desired level of isolation—thus becomes compelling. When organizations have accepted this rationale, they will move quickly to start the virtualization and consolidation process. For most organizations, this is a well-understood process that is extensively documented, and these initial deployments are nearly always associated with a formal assessment methodology that seeks to maximize the ROI of virtualization by using frameworks and best practices that govern the entire consolidation process. Unfortunately, this early commitment to following a standardized methodology often falls by the wayside, as IT and business units in general seek to  take advantage of some of the capabilities of virtualization.

A good example of the impact of this failure to follow proven processes when it comes to managing a virtual infrastructure is virtual-server sprawl. As virtualization adoption has continued to increase, virtual-server sprawl has become widely recognized as a common symptom of the management challenges that IT departments face in a virtualized environment. There are generally considered to be two forms of virtual-server sprawl; each has a distinctly different cause, although at the heart of each is a failure or inadequacy in the management processes that govern the environment.

The first form of virtual-server sprawl that is denoted by the implementation of an ever-increasing number of virtual servers without proper operational controls can be caused by two factors. Perhaps the more important of these two factors is the impact of virtualization on the nature of asset ownership within an IT organization.

Traditionally, many organizations used an asset-ownership model under which individual business units owned the servers that ran line-of-business (LOB) applications. However, to take advantage of many of the sophisticated features of virtualization—such as dynamic workload management and resource pools, and the underlying centralization of workload management—it becomes inefficient and unwieldy for ownership of server hardware to occur at the business-unit level. By allowing IT to become the de facto owner of a centralized pool of hardware assets, organizations can achieve a great deal of flexibility.

In fact, this concept of asset ownership plays a significant role in the delivery of IT as a Service. A centralized pool of hardware resources has a natural tendency toward a utility or service-provider model in which business units purchase slices of computing power and specific service levels. Perhaps more importantly, this mechanism is the only long-term approach that is both sustainable and capable of  maximizing the ROI for a virtualized infrastructure.

The second factor that contributes to virtual-server sprawl is the ease with which virtual servers can be deployed. Capabilities such as cloning and templates dramatically reduce the deployment times, when compared with a traditional physical environment. Unfortunately, this capability, when combined with a lack of well-defined and easily understood management processes, leads to a loss of management control. When it is technically easy to deploy a virtual server and business operations are no longer forced to account accurately for the costs of LOB applications, the end result will be potentially very damaging to the sustainability of an efficient and cost-effective virtual infrastructure—the upshot of which is a dramatic increase in the number of virtual machines, without any insight into the financial costs and impact of provisioning.

The second form of virtual-server sprawl comes in the form of resource sprawl, which occurs when virtual machines are provisioned with a fixed set of virtual resources (processor, memory, and storage), without due regard for the underlying workload requirement of the application. This one-size-fits-all approach often leads to underutilized hardware. In fact, this approach creates the very situation that many organizations hoped to address by implementing  virtualization.

Regardless of your particular problem (and, for many organizations, both forms of server sprawl are prevalent in their environments), a disciplined approach to IT service management can  remediate many of the issues. Specifically, in terms of the MOF, these failures can be mitigated by successfully defining the virtual machine-provisioning process as an IT service within the guidelines that the MOF specifies. By implementing all three phases (Plan, Deliver, and Operate), as well as successfully enforcing the Management layer, IT can begin to impose the necessary management processes to take control of this problem.

However, before you start, it is important to understand that the management challenges that are associated with virtualization represent a significant risk in terms of the degree of change that it brings. Because the MOF is flexible enough to allow organizations to choose the level of formality with which to approach a project, it is imperative that you acknowledge the risk that is associated with such a fundamental change to ensure that you take a very formal approach to completing each phase of the MOF. When you have developed a comprehensive framework around virtualization as a technology and how it will be implemented in your environment, you will able to approach many of the day-to-day operational tasks—such as provisioning virtual servers—less formally.

What this means in practical terms is that tasks such as deploying a virtual server can be facilitated quickly and easily—thereby, allowing IT to maintain a high degree of responsiveness to changing business needs. Perhaps more importantly, it also allows the IT organization to achieve the level of control that is needed to be successful in the long term. This tiered approach to defining and controlling the different aspects of virtualization is also critical in ensuring the long-term viability of implementing the MOF. Without the ability to streamline the requirements of the framework as they relate to daily operations, there will always be an incentive to shortcut the very processes that ensure control.

Looking at the issue of virtual-server sprawl in more depth, there are some key areas of the MOF that clearly illustrate the value, flexibility, and strength of this tiered approach. For example, consider the fact that one of the primary drivers of virtual-server sprawl is that cost of ownership is devolved from business units in such a way that business units are often no longer required to perform a financial analysis of any new initiatives—particularly, those that are in a testing or development phase. If an IT organization had used the MOF to define virtualization as a service, it would have completed the Financial Management SMF as part of the Plan phase. Because the goal of the Financial Management SMF is to ensure that there is an accurate accounting of IT expenditures, with costs mapped to specific services, the completion of this activity would have ensured that the capital and operational costs that are associated with the physical hardware that was hosting the virtual servers was a known value.

Moreover, as part of the Deliver phase, the Build SMF would have produced a design that specified an architecture for the virtual infrastructure that, among other things, would have specified the number of physical hosts and identified the computing capacity of those hosts. Combining the financial costs that were developed in the Plan phase with the capacity calculations in the Deliver phase would allow an IT organization to create a dollar cost for each virtual server that is deployed. To leverage this formal and comprehensive analysis of virtualization as a service, the IT organization would then seek to implement a much less formal but nonetheless well-documented virtual server-provisioning process. While a less formal approach to a regularly occurring task is beneficial to the overall ability of IT to be responsive, it is nonetheless an important control that will allow IT to ensure not only that virtual machines are provisioned with the correct capacity, but also that all provisioning requests follow a standardized justification process. In the long run, these requirements will ultimately rein-in server sprawl.

A specific example from the MOF that highlights how this could be achieved can be found in the Deliver phase. Using the Envision SMF and applying it to the provisioning process would allow IT to require a written justification for the proposed virtual machine. The information that is required as part of this justification could include resource requirements (CPU, Memory, and Storage), a specific project name and business-unit name, and the name of a sponsor. While this requirement could be completed by using no more than a single page—perhaps, included as part of a workflow that supported automated provisioning of the virtual machine—it would ensure that every virtual server was justified and that costs could be apportioned to the correct business unit.

Perhaps more importantly, and despite the underlying simplicity of this step, it would provide a critical checkpoint at which IT could exert and maintain control of the number of virtual servers that were being deployed and the resources that were being allocated to those servers. With these controls in place, it is possible to overcome many of the management challenges that are associated with virtualization, including the most common of all: virtual-server sprawl.

Blade servers are often cited as the cure for this problem, but they are not. Blade servers assemble server resources in a more physically efficient form

 

Conclusion

An effective management strategy is a critical requirement of any well-run IT organization. The introduction of virtualization in an environment only serves to underline this requirement, as many of the capabilities that make virtualization attractive lend themselves to  issues of control. While organizations that are new to virtualization are often unaware of these issues, in many organizations in which virtualization plays an increasingly important role, there has been a growing realization that, as a technology, virtualization brings a unique set of management challenges. Some of these challenges can be addressed through selection of the right management tools, but many of these challenges require adhesion to a management strategy that is based on a comprehensive framework that lends itself to defining the means to achieve the management goals both effectively and efficiently. The MOF provides such a guide and, when it is applied correctly, can help IT organizations impose the controls that are required to maintain a virtual infrastructure successfully.

 

About the Author

Alan Maddison is currently a Senior Consultant with Strategic Business Systems, Inc., a division of Brocade. Specializing in Microsoft  technologies, he has a 15-year track record in the IT business across many different industries. As a Consultant, he has delivered services on behalf of some of the leading IT OEMs to many companies across the country. Alan also enjoys writing about technology, and he is the author of a number of articles. You can send email to him at amaddison@sbsplanet.com.