Managing The Virtualisation 'Catch 22' : How to Combat Virtual Sprawl

Back to articles
On the Road to Cost Savings and Reduced C02 Footprint
 
Server virtualization has been heralded as a significant way for organizations to create operational efficiencies, reduce IT costs and environmental footprint. Gartner says that the effective use of virtualization can reduce server energy consumption by up to 82% , however, these savings are not a foregone conclusion. The research and advisory firm warns that IT Managers should be wary that 'virtualization can add cost and complexity' and advocates 'the use of additional (management) technology to attempt to bring the costs back into balance' . Gartner goes on to say that 'operationally, virtualization appears to be a "wash," at best — and it actually creates additional costs (people, process development and tools) on a worst-case basis' .
 
Certainly, what is true is that there is a direct correlation between the amount of savings that can be made through virtualization and how well you plan your virtualization project. You may see some immediate energy savings when virtualizing a server, but is important to be aware that although you see an initial cost benefit, spend can actually increase unless demand for those servers is carefully managed. For example, if your servers are under-utilized, non-productive, over-provisioned or your virtual server environment sprawls out of control, the potential for dwindling savings or increased spend is amplified. This is what Gartner refers to as the 'Virtualization Catch 22' .
 
The problem of virtual sprawl
Virtualization can make deployments much faster and easier. However, an increase in demand for servers can mean that your estate could sprawl out of control. In fact, "'server sprawl' is one of the biggest and most obvious causes of unnecessary energy consumption' and according to Gartner, 'may even replace physical server sprawl, resulting in server hardware (and operations) inefficiencies'. At 1E we believe that virtual sprawl must be mitigated against, so as not to negate the benefits of a virtualization project.
 
When combating virtual sprawl, organizations should consider server energy and efficiency management software that can identify when a server is doing any useful work; useful as opposed to usual or non-productive work i.e.: maintenance and monitoring tasks, such as backing up data and re-indexing. In addition, organizations should consider software tools that put their unproductive servers into lower power states and dynamically minimize energy consumption when there is no useful work being performed while constantly monitoring activity without impacting the server availability, whether virtual or physical. At 1E, we believe that such functionality (like Drowsy Server in NightWatchman Server Edition for example) provides an average saving of 12% in energy costs.
 
At 1E, we have calculated the cost of running unproductive servers to amount to $3.8 billion in energy costs alone; $24.7 billion overall, globally – there are huge savings to be made, so let's make them!
 
Beyond utilization: the CPU is not enough!
An efficient server uses most of its resources, most of the time, doing the work for which it was purchased. At 1E we believe that utilization is not the same as the productivity or usefulness of a server. We'll return to this later, but let us look for a moment at server utilization. Gartner estimates that 'most x86 server estates are running at less than 30% utilization' – with other reports citing that 'server utilization, especially in x86 environments, is often at the low end of the performance range, averaging between 7% and 15%'; an unfortunate by-product of the one server per application 'rule'. The irony here, according to 1E, is that it is these under-utilized servers have traditionally helped fuel the growth of virtualization but can result in a further Catch 22. Given that "expected server utilization rates are usually overinflated", over-provisioning may occur, which in turn leads to lowered efficiency. What's more, it is known that a typical x86 server, even at low utilization levels when idle can use almost three quarters of its total power consumption. In fact Gartner states that 'In most organizations, x86 servers are at idle use upward of 60% of their total power requirements.') Add the challenge of low efficiency to high power consumption and it is clear that the issue of server efficiency cannot be ignored.
 
Further, at 1E, we believe that measuring a server's usefulness takes more than measuring its utilization levels and that IT managers should look beyond the CPU to measure the efficiency of a server. 1E's useful work analysis reports, a key feature of NightWatchman Server Edition, empowers IT managers to configure what server activity constitutes as useful to his or her organizational needs. The reports also enable them to see when a server is doing useful work, making decisions regarding server decommissioning, virtual sprawl control, consolidation and power management simpler and with increased confidence, notwithstanding the cost and carbon footprint reductions that occur as a result of these decisions.
 
Should I start measuring the energy efficiency of my data center in 2010?
According to Gartner, 'data center and IT managers are not paying sufficient attention to the process of measuring, monitoring and modeling energy use in data centers'. Further, the research and advisory firm advises that these managers 'develop a dashboard of data center energy-efficient metrics that provide appropriate data to different levels of IT and financial management' As well as useful work analysis reports, another key component of 1E's server energy and efficiency management software is Drowsy Server functionality, which minimizes energy consumption by putting the server into a low-power – or drowsy – state. At 1E, we believe that by putting (even productive) servers into a drowsy state, this provides an average of 12% energy savings. With NightWatchman Server Edition, 1E can help IT organizations manage both physical and virtual server estate, track and report on server efficiency and energy use.
 
Cost optimization through chargeback
In cost-conscious times, chargeback schemes can help an organization gain transparency of and ultimately, optimize costs. It facilitates business and IT accountability by moving cash flow controls away from IT and back toward the lines of business, where the costs were initially incurred. Chargeback is a good mechanism to prevent or control virtual sprawl, as it helps departments or business units only use what they really need.
 
Chargeback of course relies on analyzing and recording the utilization – or usefulness – of a virtual machine. Again, in this instance, CPU utilization would not give the full picture and may man that organizations are unwittingly limiting the amount of savings that can be made. In this scenario, 1E can provide comprehensive reports, which can serve to facilitate chargeback initiatives. A key function of NightWatchman Server Edition, useful work analysis reports, give organizations insight into the cost of running their physical or virtual servers, enabling them to set smart and achievable metrics for chargeback on a per server basis. In this way, an organization can eliminate complexity and avoid having to base their chargeback framework on ambitious, technical pricing metrics.
 
Gartner recommends…
In a recent press release, Gartner states that 'Energy management can be effective only through advanced monitoring, modeling and measuring techniques and processes. Metrics form the bedrock for internal cost and efficiency programs and Gartner urges data center managers and IT organizations to make this area a high priority, which will be essential for the adoption of so many new technologies and adherence to government policies.'
 
With cap and trade carbon emissions reductions schemes either coming into force imminently or being considered by governments around the world, can you afford not to?
 
If you would like to understand how 1E can help you reduce the costs and carbon footprint within your organization, please do come and talk to us. Thank you for reading and I hope you have found this information useful.
Kind Regards,
Sumir Karayi CEO, 1E
www.1e.com
info@1e.com
 
Endnotes
i Gartner RAS Core Research Note G00165137 How to Cut Your Data Center Costs, Rakesh Kumar, 7 May 2009
ii Gartner RAS Core Research Note G00165864 Server Virtualization for x86: A Benefits Impact Assessment, Cameron Haight, Stewart Buchanan, Thomas J. Bittman, 26 May 2009
iii Gartner RAS Core Research Note G00165864 Server Virtualization for x86: A Benefits Impact Assessment, Cameron Haight, Stewart Buchanan, Thomas J. Bittman, 26 May 2009
iv Gartner RAS Core Research Note G00165864 Server Virtualization for x86: A Benefits Impact Assessment, Cameron Haight, Stewart Buchanan, Thomas J. Bittman, 26 May 2009
The "Catch-22" with virtualization is that, although it acts to improve efficiency and reduces cost in some areas, it adds cost and complexity in others, requiring the use of additional (management) technology to attempt to bring the costs back into balance. Although modest deployments of virtual servers may get along without expensive add-on management, most end users will ultimately find themselves seeking management technology to reduce sprawl, improve performance, ensure proper capacity planning, and further enable problem analysis and resolution. v Gartner RAS Core Research Note G00172255 Dataquest Insight: Server Virtualization and IT Operations Management Will Boost End-to-End Green Branding and Positioning in EMEA, Rene Millman, Mike Spink, 23 November 2009
vi Gartner RAS Core Research Note G00165864 Server Virtualization for x86: A Benefits Impact Assessment, Cameron Haight, Stewart Buchanan, Thomas J. Bittman, 26 May 2009
vii 1E calculation from ASE Energy & Efficiency report
viii Gartner RAS Core Research Note G00172255 Dataquest Insight: Server Virtualization and IT Operations Management Will Boost End-to-End Green Branding and Positioning in EMEA, Rene Millman, Mike Spink, 23 November 2009
ix Gartner RAS Core Research Note G00164493 Data Center Efficiency and Capacity: A Metric to Calculate Both, David J. Cappuccio, 18 September 2009
x Gartner RAS Core Research Note G00165864 Server Virtualization for x86: A Benefits Impact Assessment, Cameron Haight, Stewart Buchanan, Thomas J. Bittman, 26 May 2009
xi Gartner RAS Core Research Note G00164760 U.S. Data Center Conference Focuses on How to Do More With Less, Mike Chuba, Carolyn DiCenzo, Donna Scott, 6 February 2009
xii Gartner RAS Core Research Note G00172282 Q&A: Critical Issues Facing Data Center Managers, Rakesh Kumar, 29 October 2009
xiii Gartner RAS Core Research Note G00172282 Q&A: Critical Issues Facing Data Center Managers, Rakesh Kumar, 29 October 2009
xiv Gartner Says Critical Issues Facing Data Center Managers Will Worsen in 2010, 24th November 2009 - http://www.gartner.com/it/page.jsp?id=1234513
About Transend
Being environmentally sustainable makes economic sense, and Transend believes that Green IT should be a priority for every company looking to reduce costs.  Our clients reduce their impact on the environment while at the same time saving money and increasing their competitiveness in the market.
Founded in 1997, Transend achieves Green IT and environmental sustainability for our clients by recommending and supplying technology that is greener to produce, distribute and deploy, by identifying energy efficiency, consumption and waste reduction opportunities, and through environmentally sound disposal or re-use of obsolete equipment (referred to as e–waste). Transend helps clients become more sustainable with their Green IT through the supply and support of internationally leading technology solutions, such as the best in class PC and Server power management software from 1E as well as the latest developments in greener technology from IBM, Acer, Lenovo, Hewlett Packard and many more. We are New Zealand's leading Green IT specialists. www.transend.co.nz