Virtualization emerged as an area of interest in IT earlier this decade and is now unarguably one of the hottest trends in IT at the moment.  Prior posts explored the technical details of how processor virtualization and I/O virtualization function.  But what are the uses of virtualization that are driving all this attention and interest from IT, vendors and the press? 

The most important use today for virtualization is server consolidation.  During the late 1990’s and early part of this decade, small rack-mounted x86 servers proliferated rapidly.  In part, this was due to the low cost of x86 servers using Linux or Windows, compared to their UNIX predecessors.  But this was also driven by the flakiness of the software stacks on those same servers.  The software and operating systems were not nearly as robust as they are today and it was problematic enough that a lot of IT staff considered isolating applications from one another a best practice – even if it lead to low utilization.  There were plenty of peculiar bugs that resulted from a combination of corner cases in the OS and different software packages and using a server dedicated to a specific application avoids that problem entirely.

The proliferation of small, underutilized servers was slowed down by the dot-com bust in 2001, but also by constraints on data center space, cooling and power.  Cooling and power were particularly sensitive, because while Moore’s Law means that you can get more computing power in a chip over time – it also means that the power density (whether it is W/cm² or W/cm³) increases substantially.  So a data center designed for 1995 servers is wholly inadequate for those in 2003.  Worse yet, data centers are incredibly expensive to build or redesign – for many companies there is a bit of pain as your data center approaches 80% capacity and the cooling or power system gets stretched.  But typically, building another data center is far too painful and expensive to be an acceptable solution.

In 2002, VMware’s first server products came out and started the trend towards virtualization by providing the isolation that IT departments needed, but within the context of a single server.  From there, it became a simple equation as server consolidation reduced power costs, freed up data center space and reduced future expenditures on servers.  The overall poor economic environment is accelerating the trend towards virtualization as the cost of a virtualization project can easily be outweighed by the savings over a short period of time, thus enhancing the bottom line.

But above and beyond consolidating multiple physical servers onto a single server, where can virtualization go?   Snapshots and rollbacks are a spectacularly useful feature, even for an individual user, as opposed to a corporate IT department.  Since virtualization maintains all the state for each instance separately, it is very easy to save a copy of the state of a virtual machine at a given point in time, creating a snapshot.  At any point later in time, the user can rollback the VM to the state recorded by the snapshot, which has the practical effect of creating an ‘undo’ button for anything on a PC.  This is equally handy to corporate IT departments, which can create snapshots before making any substantial changes.

Migrating a virtual machine is a close cousin of the rollback technique described above.  Instead of rolling back the same VM to a prior state, the user can send the state to a different VM, on a different physical computer system even.  Often when a VM is migrated, it can be done ‘live’, without noticeably interrupting service; a paper from the research group behind the Xen hypervisor documented migrating a Quake server with under 60ms of downtime.  This is a boon to IT staff, who can use live migration for load balancing by moving VMs from an over-crowded server to a new or underutilized server. 

Another key use for virtualization is providing backwards compatibility for older applications or operating systems.  I can still vividly remember my first job in 2000, where I learned that the engineers were all still using Windows 95 because one of their engineering applications was still DOS only!  With virtualization that is no problem at all – and you don’t get stuck with an ancient and defunct OS.  Apple popularized the use of virtualization with binary translation to enable backwards compability for older PowerPC applications on the newer x86-based Macs, via Rosetta and later Boot Camp.

The last application for virtualization is testing and validation.  For many software or hardware companies, testing to ensure compatibility is a major undertaking.  Before the advent of virtualization, systems had to be configured with different combinations of real test hardware and software.  With virtualization, dozens of configurations can be tested on a single server simultaneously.

While a catalogue of the various uses of virtualization is interesting and provides some insight to the value that virtualization can provide, the implications for server design are in some cases more fascinating, and that will be the subject of my next post.  

Unless you haven’t picked up a business magazine in the past eight years, chances are you’ve at least heard of Enterprise Risk Management. And if you’re reading this, there’s a good chance that you’re intimately familiar with it.

Not to be confused with that other ERM familiar to IT types (Enterprise Resource Management), risk management falls under the umbrella of yet another acronym that’s reached buzzword-status among IT managers: GRC, short for Governance, Risk and Compliance. After all, what has taken down some of the greatest names in business recently? Lousy governance, failure to predict risk and sloppy adherence to regulatory requirements.

Proactively and prudently managing these challenges, which include an assortment of technology-related risks, has become part of the financial responsibilities that make up fiduciary duty.

What’s more, scenarios that several years ago may have seemed outlandish to most businesses – a terrorist attack or a calamitous natural event — now warrant straight-faced consideration. In a post-9/11 world, overlaid with threats of rapidly shifting climate patterns, it seems that nothing can be ruled out.

As a result, virtualization merits consideration under any comprehensive ERM strategy. For disaster recovery, for instance, it’s a no-brainer. And virtualization’s distributed nature also offers some advantages in scalability, dramatically reduced IT capital and operating costs, along with many other efficiency initiatives.

But virtualization has potential drawbacks, too, most notably a unique and well-documented set of security challenges, but also the following:

 

1.  The virtual layer expands more rapidly than physical IT assets, making it hard for security to scale up proportionately (even as other functions become easier to scale up).
2. Virtual servers can “disappear” during periods of disuse, only to reappear when needed, but they may have missed crucial security updates in the interim.
3.  Attacks on the software supporting virtualization and virtual machine proliferation raise risks that don’t exist in the solely physical server world.

 

Despite these concerns, adoption of virtualized solutions appears to be headed nowhere but up. One measure: Virtualization leader VMware, which releases its fiscal year 2008 earnings later this month, had a remarkable 3^rd quarter in 2008 — with sales up 32%, to $472 million.

So, how much should companies worry about low-probability events — those disaster recovery scenarios with the potential to wipe out a data center, such as natural catastrophe or terrorist attack?

They should at least get realistic probability assessments and make decisions based on those assessments, say Wharton professors Howard Kunreuther and Michael Useem. The problem is, it often doesn’t even get to the assessment phase, the professors say. Kunreuther is co-director of Wharton’s Risk Management and Decision Processes Center. Useem is director of Wharton’s Center for Leadership and Change Management.

 

“Most executives tend to avoid thinking about low-probability events until after they occur,” the professors wrote in a December op-ed piece that appeared in The Philadelphia Inquirer. “They fall into a trap of believing such events will not take place — at least, not on their watch. The implicit principle is ‘NIMTOF’: Not in My Term of Office.” And yet, “The art of leadership is anticipating the unpredictable,” they write.

In the corporate context, one remedy to “NIMTOF” is effective governance, according to Kunreuther and Useem. “Good governance implies appointing forward-thinking directors with long-term perspective, who will guard against low-probability, high-consequence events.”

There’s no question, virtualization has matured nicely over the past few years. Recently, with data center battle-tested products like VMware, it has become a key enabling technology for an even broader scope of markets. For the data center and enterprise, the benefits of virtualization are numerous and obvious. The need for high availability platforms that scale on demand has paved the way for larger, application-aware and multiple OS capable architectures. In addition, server consolidation to provide efficiencies in power consumption, maintenance and other overhead costs, has become critical. There are lots of other areas where virtualization reduces costs and provides efficiencies, including cooling, application/OS testing and associated man hours, as well as reduced backup, security and OS software licensing fees. For many in the enterprise, virtualization is a virtual no-brainer. In fact, many current business models in IT wouldn’t even exist without virtualization today.

On the other end of the spectrum, the mainstream consumer or small business has been living in a “maintain and upgrade every 2-3 years” paradigm for a very long time for their generalized computing requirements. It has been only recently that the average consumer end-user has seen the benefits of virtualization technology but they are becoming more apparent here a well. Though perhaps it’s not the be-all, end-all of general computing, virtualization has shown its merits to an ever-increasing base of end-user types. Though some may claim that cloud computing and virtualization are different, there are many commonalities between what the enterprise and data center markets call virtualization and what end users have at their disposal now for online application, backup, and synching services. From MobileMe to Amazon’s EC2, virtualization has now officially gone mainstream and there’s no end in sight with its numerous application potential and extremely low cost model. Though the computing enthusiast or gadget freak may not be comfortable with a reality where all of their processing and storage resources are handled virtually, let’s face, the mainstream end user simply doesn’t have much use for all that hardware.

A few years from now, many end users will be comfortable with a simple netbook or a thin client as their desktop and then the rest of all that technology will reside in the cloud. In short, from the data center to the enterprise and now the end user, virtualization is here to stay and it’s not just for CIOs and Senior Technicians anymore.

Is there something IT professionals can do that has the potential of addressing this triple threat of social, economic and environmental ills? The answer could be virtualization. Proven technologies that allow a single server to take on a variety of unrelated tasks that would normally require the use of multiple dedicated servers offer some obvious benefits in the areas of cost savings and efficiency. Can those benefits also translate into improved fiscal performance, reduced consumption of energy resources and a lower carbon footprint? The consensus seems to be in the affirmative.

In fact, these are among the reasons Gartner Group recently anointed virtualization as the No. 1 strategic technology for 2009, topping trendier tech topics such as Business Intelligence (#2), Cloud Computing (#3), Green IT (#4), and Unified Communications (#5). In his blog on the unveiling of this year’s Top 10 list, Gartner’s David Cearly cited virtualization’s “significant potential to reduce IT costs” as the primary business driver, noting that widespread IT belt-tightening in 2009 would prompt popular server virtualization technologies to be augmented by an uptick in virtualization techniques applied to storage and client devices.

Next to the promise of cost reduction, energy savings is among the biggest perceived benefits of virtualization. A study performed last spring by independent Washington, D.C.-based research firm KRC Research asked 205 IT decision makers in the retail industry to name their reasons for deploying virtualization in their data centers. Thirty-one percent pointed to energy savings as a major factor in their decision. According to the authors of the report, “The rising awareness of ‘green’ energy initiatives, as well as the increasing power required to run and cool data centers, likely factored into this response.”

So can making each server do the work of several actually reduce a business’ impact on the environment, as well? According to a recent McKinsey report, data centers contribute a substantial amount of greenhouse gases – potentially climbing to a whopping 1.54 metric gigatons by the year 2020, which is 3 percent of all emissions. “The fastest-increasing contributor to emissions will be growth in the number and size of data centers, whose carbon footprint will rise more than fivefold between 2002 and 2020 as organizations in all sectors add servers to meet rising demand,” the report states. Virtualization is one of the primary tools businesses have for reducing data center sprawl and its accompanying carbon footprint.

Given the current economic climate, as well as the new priorities that are expected to be ushered in by the next administration, companies that are “lean, mean and green” will be the most likely to survive and prosper in the year ahead. Organizations that have not yet deployed virtualization need to take a long, hard look at the technology and assess what it can do — not only for their bottom line, but for their country, and even their world.