Virtualization Technology News and Information
VMblog's Expert Interviews: Jeff Klaus of Intel Discusses Green Data Center Initiatives


Is something as simple as constantly checking Twitter and Facebook causing damage to the environment?  The more we connect to the Web, the bigger the job for data centers; and the bigger the job for data centers, the more energy is needed to operate them.  To discuss how green initiatives are changing how data centers operate, and how they can allow us to connect to the Web guilt free, I reached out to an expert in the field, Jeff Klaus, GM for Data Center Solutions at Intel, to help answer the question and explain how recent green data center initiatives can save money and increase energy efficiency. 

VMblog:  We're hearing a lot more these days about green data center initiatives, can you define what we're talking about here for folks who may not know?

Jeff Klaus:  Operating a green datacenter typically consists of at least two of the following objectives. 1) Get the work done as efficiently as possible - which by definition means extracting the maximum performance for the minimum resource and energy consumption, and 2) Taking additional initiatives to meet corporate social responsibility (CSR) goals - such as sourcing a certain percentage of power from renewable resources.  Every company will make their own choices. These may be perfectly compatible with the goal of maximizing performance while minimizing energy consumption, or a company may proportionately pay more for higher priced infrastructure because of its operational efficiency.

VMblog:  What are 3 steps data center operators can take to start implementing green datacenter / energy-efficient initiatives?

Klaus:  Firstly measure; we typically say "If its worth doing, its worth measuring." Operators need to determine what they are dealing with today, before proceeding to some new state, so they can see the benefit based on their actions.  Use or develop relevant KPIs - Key performance indicators to identify what is most important to their business.

Relative to infrastructure and IT efficiency, operators should consider a DCIM solution on the market that supports many types of capabilities and usages.  But their main goal is to reduce cost, improve resource utilization, and provide ongoing analysis tools.

Lastly, operators need to develop reasonable targets to achieve.  One example could be to. increase total performance by 20%, but maintain the same operational budget or reduce overall operational expense by 20%.   They should trial solutions and evaluate them relative to these goals to assess the value of implementing them.

VMblog:  Modern day, x86 server virtualization goes back to around 2000, and the major use case that caused it to take off around 2003-2004 was consolidation. Now we're investing in virtualization and cloud technologies.  How have these things helped the green data center initiatives?

Klaus:  Despite the technical differences between virtualization and cloud technologies, there is commonality in the goals and the means.  Virtualization definitely benefited end users by facilitating, through consolidation, higher utilization and lower cost per compute by more efficiently loading servers. One way to view cloud usage is that it extends this thinking with the additional benefit from scaling out.  That is Cloud Service Providers (CSPs) are in the position to deliver their respective services, by reducing IT overhead, application specialization and competitive business focus - as they operate at scale for many customers. So it is perfectly feasible to see this as a continuum with some IT services provided in-house and some more standardized offering provided by CSPs.

VMblog:  I've always been told that server virtualization is a leader in green data center initiatives, and it's been around for more than a decade.  Is it not enough?  Has it not solved the problem?  Has it only put a dent in the problem?  Have things shifted further?

Klaus:  Virtualization has provided significant benefits to datacenter operators.  To start, hardware utilization has improved and efficiency is increased.

Further virtualization and related consolidation may be inhibited by more difficult application performance issues, real business cost ,or risk to deploy further given the effort required.  How many native single thread applications are running because no one wants to attempt to migrate because of the risk to the underlying business?

VMblog:  How can companies revamp their current centers to be more efficient, instead of building new data centers?

Klaus:  As mentioned before, measure - determine what you dealing with now, before proceeding to some new state to determine any benefit. 

Operators should also use their existing KPI - Key performance indicators or refine new ones to help identify what is most important. This will vary based on the nature of the industry, but could include answering some of the following:

a.  How long does it take to deploy a new application?

b.  What is the operational cost of making an application upgrade?

Remove unnecessary infrastructure - Virtualization will help find some orphaned servers, but this is an ongoing task to ensure that servers that are deployed are being allocated and utilized - by their respective lines of business. 

Lastly, research and short-list best practices that may be feasible within the constraints of your datacenter, then deploy, measure, correct, and repeat along the way.  You can consult The Green Grid Data Center Maturity Model as a guideline.  On the facility side - typical areas to start are deployment of cold aisle/ hot aisle containment, use of free air-cooling, HVAC with control logic.  On the IT side, upgraded servers in conjunction with virtualization can dramatically reduce the overall power footprint.

VMblog:  What metrics can you point to that justify green-initiatives?

Klaus:  The Green Grid developed and published PUE which provides a metric to describe the overall efficiency of IT equipment to the facility.  This helps provide a metric for comparison against other datacenters. You need to use the tool as a yardstick for performance, but understand the trade-off in power efficiency on both the IT side, as well as the facility side of the equation, so that you do not inadvertently create a ‘better' PUE, at the expense of real efficiency gains. Additional efficiency metrics are in development within the Green Grid allow for improvement on more specific vectors - such as water use (WUE). 

There are also some immediate actions that operators can take to reduce cost and energy use, including:

i.     Boosting airflow management can reduce energy 40%

ii.     Consolidating servers can reduce energy 10-40%

iii.     Improving processing technology increases computer efficiency 6 fold

iv.     Innovative cooling technologies can reduce energy up to 95%

v.     Raising the temp. can decrease cooling costs 60%


I'd like to thank Jeff Klaus, GM of Data Center Solutions at Intel, for taking time out to help educate us on recent green data center initiatives.

Published Tuesday, February 16, 2016 6:37 AM by David Marshall
Filed under: ,
VMblog Video Q&A: Intel GM Jeff Klaus Discusses the Future of the Green Data Center : @VMblog - (Author's Link) - March 7, 2016 8:16 AM
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<February 2016>