Virtualized environments crying out for capacity management

By Drew Robb

Virtualization is one of the most rapidly adopted technologies of all time. From nowhere a few years ago, there will be an estimated four million Virtual Machines (VMs) deployed worldwide within a year. And the advantages are many – higher utilization, rapid deployment of servers and reduced data center footprint, to name a few.

While users love the flexibility this brings, the sheer number of VMs now in play has made it painfully clear that unmanaged virtualization creates problems such as administrative complexity and resource unpredictability. 

“Organizations are quickly realizing that virtualization without proper planning, management and service optimization capabilities can lead to performance headaches that can stymie attempts to leverage the technology for real business advantage,” said John Madden, an analyst for Ovum Research.

Unchecked growth of virtualization in the enterprise, then, is creating serious challenges. Many of these issues are solved by the skillful application of capacity planning and performance management. To achieve that, however, capacity planners must think differently about such environments and must deploy tools that are able to address the complexity of virtual architectures.

“TeamQuest has focused its products particularly on working with organizations to help them optimize virtualized systems,” said Madden. “TeamQuest’s portfolio of IT ser vice optimization products includes its TeamQuest Model capacity planning tool to help IT organizations plan for forecasted and unforecasted business requirements, configure and determine the number of virtual machines needed, and determine when servers are likely to reach capacity.”

The Lure of Virtualization

According to research firm TheInfoPro (TIP), over 50% of enterprises use virtualization today. That rate of adoption is far greater than those of other technologies TIP has been tracking for many years. Clearly, this technology has become an integral part of the IT landscape.

So what is virtualization? In essence, it creates an abstraction layer between a physical machine and the managing application. VMware ESX Server, for example, allows companies to run multiple VMs within one physical server.

“Rather than presenting resources in a way dictated by their implementation, geographic location, or physical packaging, virtualization provides a logical rather than physical view of data, computing power and storage capacity,” said Per Bauer, POSITION, at TeamQuest. “An abstraction layer decouples the physical hardware from the operating system to deliver greater IT resource utilization and flexibility.”

Service deployment, for instance, can now be done far more rapidly. A decade ago, it took several weeks to order and provision a server whereas today it can be done with a VM within the hour. Virtualization also hides complexity from users and applications. As it greatly facilitates server consolidation, virtualization makes it possible to cut down on the amount of ventilation, electricity, cooling, staff and space required per unit of processing power.

VM Sprawl

Unfortunately, the honeymoon appears to be over between organizations and virtualization. Indiscriminate rollout of VMs is now creating headaches in many areas. According to a survey by Ovum Research, 40% of mid-sized and large enterprises cited bottlenecks and other performance issues as the top challenge with virtual systems. Proper sizing of VMs and unexpected workload spikes were experienced by 33% of end users. 25% were concerned about how to recover from performance problems in virtualized systems. 

“VMs exert a performance overhead compared to physical servers, though we expect this to diminish over time,” said Tim Grieser, program vice president for the Enterprise System Management Software group at International Data Corp. (IDC). “In addition, some workloads don’t run well in such an environment, and network traffic can degrade under virtualization.”
While virtualization shelters complexity from the users, it adds complexity to administration. And that means expensive, highly skilled personnel are required more than ever. Proprietary differences are also apparent. Many vendors advocate their own virtualization schemes and the variant approaches can be quite confusing. This factor makes it hard to compare apples to apples across systems. While it is a simple matter to measure utilization rates in a physical CPU, for example, it is a real challenge when that processor is virtualized among 10 VMs. Further, with processors being dynamically added and subtracted on a regular basis, licensing becomes a real issue. If you are swapping out applications on VMs, how many licenses should you pay for?

Perhaps the biggest problem, though, is unchecked VM expansion. Companies are slowly waking up to the fact that physical server sprawl is being replaced by its virtual equivalent i.e. it is so easy to deploy a VM that most organizations have no real idea how many they have, what physical server they are linked to and how many resources each requires for optimum usage. 

“Proliferation of VMs starts immediately and rapidly becomes VM sprawl,” said Grieser. “Thus many companies are growing concerned about how to manage a virtual infrastructure.”

The Vital Necessity for Capacity Management

Capacity management provides a point of focus and management for all capacity and performance related issues. It facilitates the process of matching the capacity requirements of IT to existing and forecasted business demands. This is particularly important in virtualized environments where VMs proliferate on a daily basis.

“The beauty of capacity management is that it allows you to do what-if analysis for use-case scenarios, determine the impact of changes on system performance and build mathematical models to calculate the impact of changes,” said Grieser. “It is invaluable in supporting ongoing sizing and provisioning initiatives.” 

But traditional capacity management methodologies may not perform well in the virtual world. With so many different virtualization schemes available, it’s no wonder that some capacity planning metrics have become inconsistent. Let’s look at a real world example – an IBM micropartition in an AIX 5.3 simultaneous multi-threaded (SMT) environment, for example, might consist of 2 virtual CPUs in a shared processor pool. When you come to measure such an environment, it presents some challenges. The results can be different if SMT is enabled or disabled, and if the processor is capped or uncapped. The bottom line is that the addition of virtualization means that CPU statistics can differ tremendously depending on which one is used.

So an analyst might look at one metric and think utilization is low whereas another takes a different metric and sees an entirely different picture. So who’s right? It’s not an easy question to answer with virtualized processors. Each answer is correct from its own perspective. You have to understand what a specific metric reflects and how it relates to the overall picture.

This is just one example of the confusions that roll out of VMs can create. Not surprisingly, the management capabilities associated with virtualization have increased markedly in recent years. VMware VirtualCenter, for example, rapidly provisions virtual machines and monitors the performance of those VMs as well as physical servers. VMware VMotion leverages the virtualization of servers, storage and networking to move an entire running virtual machine instantaneously from one server to another. Distributed Resource Scheduler automatically balances the workload between pools of virtual servers, allowing servers to operate at eighty percent or greater utilization. In addition, server vendors have released management tools to reduce the burden of virtualized server operation. 

“While such products help manage environments, they don’t tell you what you need, don’t understand business goals and don’t help you look into the future,” said Bauer. “The bottom line is that there is no substitute for capacity management.”

TeamQuest Addresses the Virtual World

Virtualization, therefore, needs capacity planning more than ever and the TeamQuest Performance Software Suite is uniquely addressed to virtual environments. It answers questions such as

• What happens to response times if another five VMs are added?
• Where is the company experiencing over- and under-capacity?
• When will the application load exceed physical server capacity?

TeamQuest lets IT plan capacity effectively, even in the most complex virtualized architectures. Take the case of a large insurance firm that demanded a response time of 1.5 seconds for a new application. Using TeamQuest Model, the capacity planner discovered that this solution would cost $15 million. Further modeling revealed that a 3-second response time would reduce the budget to $12 million and a 5-second response time would cost $10 million. By providing management with this information, they took a second look at their original specifications, concluding that it was better to accept a short delay than to pay an extra $3 million for fast response.

When it comes to consolidating servers, capacity planning can help you find the right candidates, configure them optimally and consolidate with confidence. One firm used TeamQuest to save $15 million the first year, $7.5million the second year and a further $15 million during the third year. In this case, capacity planning was used to prove that service levels could still be met for each consolidation project. Thus capacity planning makes consolidation projects feasible and makes the business more efficient.

Capacity Management More Important Than Ever

Server management and other VM technologies will never eliminate the need for capacity management. These tools certainly add value but can’t do the whole job. They don’t understand business priorities and can’t look into the future. Thus capacity planning is more important than it has ever been. It aligns IT resources with business priorities, provides choices for business decision makers and offers a bridge between business units and IT operations.

With VMware and other virtualization technologies being so prevalent, however, capacity planners must become familiar with the technology and become experts in how it influences capacity management. TeamQuest will prove invaluable in this regard. It answers questions such as what happens to response times when VMs are moved from one server to another, when will a specific guest become “full” and when will application activity on all VMs exceed server capacity. It provides tools to determine current capacity, view ongoing capacity requirements and plan for the future using what-if scenarios.

“Capacity planning tools such as TeamQuest play an important role in cost containment within a virtualized environment, as well as demonstrating ROI,” said Grieser. “TeamQuest enables you to look at the physical server and also drill down into the individual VM to see graphically how it is performing.”