IBM: Leading the Way in Virtualization for Almost 40 Years

Virtualization is the new frontier for x86 servers, providing customers with better utilization of their server assets, simplification of their management, and a reduction in their overall costs. But many issues may confront them as they deploy this new technology. People have adopted virtualization for many different reasons, but one clear fact remains: the implementation of virtualization is as important as the benefits it provides.

The questions are legion: What degree of virtualization can I expect to achieve? To what extent can I consolidate servers? How much will I save on power and staff? Will system performance be at risk during peak periods? Is the present system configuration optimized for virtualization? These are just some of the questions racing through the minds of managers in organizations on the verge of adopting virtualization.

Businesses want to "get it right" from the start and have no patience for false starts and drastically revised project scopes. After all, virtualization is a solution to the problems of server sprawl, complexity, and inefficient processor utilization, as well as staff resources, and is not supposed to be another cause of sleepless nights.

Embarking on virtualization need not be a voyage into the darkness. Getting it right starts with the selection of a partner that can help you develop a reliable roadmap and realistic timetable for virtualization. Clearly, that partner should have a depth of expertise in virtualization, a breadth of experience with organizations like yours, and a track record of successful on-time, on-budget implementations.

A Good Plan Begins with a Reliable History
A clear-eyed guide is at hand. As a virtualization partner, IBM is unrivaled in its experience and expertise. IBM pioneered virtualization with VM/370 on the 370 mainframe computers nearly 40 years ago. At that time, mainframes were the workhorses of the data center. As demands on businesses grew, companies were required to run multiple tasks on mainframes to meet the demand. However, online and batch workloads couldn't easily coexist on mainframes, and organizations were understandably reluctant to invest in a second mainframe. The VM/370 operating system allowed multiple environments to run, which solved their problems by enabling online and batch workloads to be processed simultaneously.

With the advent of low-cost PC servers, the need for virtualization temporarily lost its urgency in most organizations. Yet, some of the same early challenges remained. Applications on servers were single-thread, and processors couldn't handle multiple workloads simultaneously. However, this fact posed a less serious hurdle in a PC environment. Since PC servers are far less expensive than mainframes, data centers responded to requirements for new operating systems or time-sensitive workloads by simply buying another server.

"IBM's Sizing Guide starts with a careful analysis of both the source servers and the target servers, both in terms of current and projected loads. The data gathered is then analyzed, and a plan is developed to avoid artificial "hot spots," or surges in workload demand that can slow down response times."

Fast forward to today's data center. Gradually, it is dawning on business leaders and IT managers that the practice of adding servers as well as storage devices at random has meant that massive amounts of idle processing capacity has crept into their data centers. Not only that, but server sprawl has strained power supplies, increased cooling costs, required greater technical resources to manage a complex environment, and pushed space requirements to the limits. Fortunately, virtualization is now possible on Intel- and AMD-based servers. Instead of needing four servers to run four applications simultaneously, a small business now needs only one server with virtualization to run the same four applications, thanks to virtualization.

Once again, virtualization is a hot topic, and IBM is the trusted partner of choice. Recently, the publication Information Week named virtualization as one of the "Five Disruptive Technologies to Watch in 2007" (link resides outside of ibm.com), adding, "The cost savings promised by virtualization will be too compelling to pass up."

IBM's Sizing Guides Help Determine the Plan for You
Since virtualization became an opportunity for server farms in recent years, IBM has been in the vanguard of the trend. In the past five years, IBM has been performing server consolidation studies for their customers using tools like our Server Consolidation toolkit. The data and insight gained from these studies was used to produce the IBM Virtualization Sizing Guide. The sizing guide is derived from real world usage, and therefore provides a guide that can be used to help implement virtualization in real situations. IBM's past experience can be applied to today's real world workloads and provide guidance to its customer on how to implement virtualization. This guide helps organizations to build a better virtualized infrastructure. In 2006 alone, IBM conducted 389 server consolidation studies, a 46% increase over the prior year.

IBM's Sizing Guide starts with a careful analysis (using VMware Capacity Planner and the IBM Consolidation, Discovery & Analysis) of both the source servers and the target servers, both in terms of current and projected loads. What are the average and peak utilization rates for both processors and memories? What are the disk and network input/output rates? What quality of service standards do you want to maintain? This data is then compared with similar studies done for 30 client organizations.

The data gathered is then analyzed, and a plan is developed to avoid artificial "hot spots," or surges in workload demand that can slow down response times. In situations where a business lacks clarity on the future resource needs of some workloads, a scalable server, such as an x3950, might be recommended so that additional workload can be handled without new server hardware or other physical changes.

As you might expect, IBM's wealth of experience with successful implementations of virtualization has led to some valuable insights that will benefit clients. Here are but a few insights gleaned from years of successful implementations:

• Variations in processor utilization will determine the number of virtual machines possible on a server configuration.
• To calculate the amount of headroom or excess capacity needed, focus on peak periods of operation rather than average operating periods.
• To correctly size the number of virtual machines per platform, consider average and peak utilizations of processor, memory, and disk.
• Physical processor utilization is not always the primary constraint for consolidated workloads.
• Platform selection is driven by size and number of virtual machines.

As a trusted partner, clients can take comfort in the fact that IBM is actively involved in standards-setting bodies for virtualization. This insures that virtualization remains self-contained and hardware agnostic, so that clients won't one day find their investment in virtualization suddenly obsolete. Clients can also be assured that engaging IBM won't mean a "rip and replace" upgrade to their existing data center.

IBM is committed to work with clients to redeploy their existing systems. As a rule of thumb, the past two generations of servers usually can be redeployed in a virtualized environment. Moreover, legacy software can be transferred quickly and easily to the virtualized environment, something that previously required days or even months to occur.

Simplicity, flexibility, and efficiency are the siren calls of virtualization. Luckily, IBM is there to show you the way.