Trends in data center power was the topic at AFCOM New England Chapter’s meeting this week, and apparently it’s a subject that resonates with members—at least judging by the nearly 100 attendees who showed up. (As the New England chapter enters its third year, President Rocko Graziano, whose real job is manager of infrastructure operations and services at L.L. Bean, said this was the largest meeting yet). Two speakers gave the audience their take on the some emerging trends they see taking shape.

Rudy Kraus, CEO of Validus DC Systems, a provider of direct current (DC) power infrastructure for data centers and telecommunications facilities, naturally sees a bright future for data centers powered by DC rather than AC-based electricity. Kraus cited a number of statistics from the likes of the Uptime Institute and McKinsey outlining just how much power data centers can save by switching to more efficient DC power. If data centers in the United States converted only 10% of their capacity to DC power, that would eliminate $1 billion in electric bills. The co2 emissions for a 10 megawatt data center with 17,500 servers would drop from 99,776,400 pounds to 59,865,840 pounds. Kraus invited members of the audience to do their own comparison by visiting an online calculator offered by Intel that analyzes facility-level efficiency of AC and DC servers.

The other speaker, Brian Ouellette, of J.S. Fleming Associates, a provider of power and cooling systems, spoke about the five power trends heading to a data center near you. The top trend, that energy efficiency is gaining importance, is pretty self-evident. The other four trends centered on ways to make data centers more efficient: New ways to scale UPS architectures into adaptive models that can adjust to changing power requirements; two-stage power distribution that reduces restrictions to cooling air flow, among other benefits; increasing use of monitoring with tools such as smart power strips (that monitor in-rack power) and branch circuit monitoring (that monitor each PDU output circuit). Ouellette also pointed out that data centers don’t have to go high-tech in order to become more efficient. When Ouellette asked the audience whether they use blanking panels in their data centers, only five people raised their hands. “Blanking panels are a great way to get the air where you need it,” he said. “Otherwise, you’ll get cross-contamination of air from your hot aisle and cold aisle.”

Google seems to be everywhere. But where does the omnipresent company actually exist? This topic has gained a lot of attention lately from this Google Data Center FAQ to the subsequent map put together by the folks at Pingdom. But why is the location of the super-secret-yet-ubiquitous Google getting such a large amount of attention?

Google’s data centers have had a lot of attention around the country. In Oregon, initial secrecy surrounding the company’s development of The Dalles facility was an object of a lot of public discussion, which resulted in the facility earning the moniker of “Voldemort” (a reference to the fact that the Harry Potter series character is most commonly referred to as “He Who Must Not Be Named,” and that local officials couldn’t say the “G” word while the facility was under construction). But Google let up a bit and put on a more public face in 2007, including a site tour by local reporters from The Dalles Chronicle. Google had changed its tune on keeping the activities on site “super secret” after a public backlash concerning the 15-year tax break the company received from the State of Oregon, and power consumption concerns. Indeed, the energy consumption of the company’s data centers has attracted national media attention, beyond IT-focused outlets. As the company has grown in power over the past ten years, it has increasingly attracted attention, first with its quirky dot-com work environments and employees (including Google bean-bag chairs that grace the Googleplex), to controversy over acceptance of Chinese government restrictions over search-engine deployment.

But the company hasn’t turned to total transparency with all of their locations, and thus, members of the IT community still get excited when they can learn one more “secret” about where the company actually exists.

So the question that’s nagging me is, what can everyone else learn from Google’s experience? Secrecy is good for business, but up until what point? I come to TechTarget with a background in journalism and public affairs, so this question is not unfamiliar to me. Many companies rely on keeping their activities and intellectual property secured, it’s common in business. But, is Google being too extreme? Do their data center facility locations really need to be kept on the “down-low”? And why do datacenter insiders care?

I asked this question to Robert Cringley, of I, Cringley. And he gave me some interesting insight into the big-picture as it were.

“Google wants a peering relationship with every broadband ISP of substance,” explained Cringley. So, by having data centers located in proximity to ISPs, the ISPs and end-users benefit through high-speed connections at low cost. However, the self-serving part is that this relationship makes everyone more dependant on Google for everything (storage, search, carriage, etc.). “It’s all aimed at the ultimate replacement by Google of current audio and video distribution networks. Look 5-10 years out. That’s why geographic distribution is so important, because they really need to have a copy of the Google data set one hop from every broadband ISP.”

What do you think?

Sydney, Australia based-Ideas International Inc. has launched an open source-style website to compare and rate the functional capabilities of blade servers on Monday, April 7.

The IT research and analysis company’s new site for Collaborative Product Evaluation looks at medium-sized blade servers and will include enterprise-level blade server data by mid-summer, said Jim Burton, the vice president and senior analyst for entry-level servers and blades at Ideas International.

The site lets users compare various components of the servers that fall under the umbrellas of platform functionality, environmental footprint, virtualization functions, reliability, serviceability and manageability, and deployment considerations.

The information is based on the hardware specifications, interviews with end users, and performance data, Burton said.

“We establish the appropriate ratings, but it is an open source-style website, so users can affect these ratings too,” Burton said. Of course, Ideas International give the user feedback a credibility rating, so only statements supported by concrete data can actually bring a rating up or down, he said.

The site is pretty handy if you are on the market for blade servers, especially because the site allows you to make comparisons based on your priorities. If you need power efficiency, you can compare boxes based on that alone. Same goes for factors like “green-ness,” cost, networking and so forth, said Burton.

Ideas International also has evaluation sites for x86 virtual machine platforms and plans to create evaluation sites for Unix-based systems and Linux in the near future, so keep an eye out for those.

MTechnology, a research firm that runs probabilistic risk assessment analyses of electric power sources, has released a study finding that flywheel-based uninterruptible power supply (UPS) systems are more reliable than the more common battery-based UPS during short outages of 10 seconds or less.

The study was paid for by Active Power, a flywheel-based UPS company, so you can take the conclusions for what you think they’re worth. Martin Olsen, the product manager and development director at Active Power, said that they “made it very clear that (MTechnology) would be doing the study, and to make it as objective as possible. That’s what they do for a living.”

Flywheel-based UPS systems have gained some traction in data centers as an alternative or complement to battery-based UPS. Terremark, a large data center colocation company, just recently installed flywheels instead of batteries at a new campus in Virginia. How it works is the primary utility power gets the mechanical flywheel spinning. When the utility power cuts out, inertia keeps the flywheel spinning at thousands of rotations per minute. That rotation is harnessed into power that can support the data center for short periods of time until either the utility power returns or the switch is made to backup generators.

The MTechnology study compared a flywheel UPS to double-conversion battery UPS and found that the flywheel was almost seven times more reliable than the battery UPS during short outages (10 seconds or less). During longer outages, failure rates were about the same.

Intel Corp. introduced two low-voltage 45 nanometer (nm) quad-core Xeon processors today that run at 50 watts, or 12.5 watts per core, and frequencies up to 2.50 GigaHertz (GHz).

The Intel Xeon Processor L5400 Series is built on 45nm manufacturing and a transistor formula that, combined, boosts performance and reduces power consumption.

The Quad-Core Intel Xeon L5400 processors are as much as 25% faster and have a 50% larger cache size than Intel’s previous-generation low-voltage Quad-Core Intel Xeon processors introduced last March, while maintaining the 50-watt thermal envelope. The quad-core L5420 and L5410 processors run at 2.50 GHz and 2.33 GHz, respectively, and have 12 megabytes (MB) of on-die cache and dedicated 1333 MHz front side buses (FSB).

Vendors supporting the L5400 and L5210 series include Asus, Dell, Fujitsu, Fujitsu-Siemens, Gigabyte, HP, Hitachi, IBM, Microstar, NEC, Quanta, Rackable, Supermicro, Tyan and Verari.

Next quarter, Intel will also begin shipping a new dual-core low-voltage processor with a 40-watt rating and clock speed of 3 GHz, with a 6 MB cache size and a 1333 MHz FSB.

The Intel Quad-Core Xeon processor L5420 is $380 per 1,000 units and the Intel Quad-Core L5410 is priced at $320 per 1,000 units.

After encountering each other on numerous customer engagements, Hyperic Inc., an open source systems management company, and The OpenNMS Group, an open source network management company, are officially forming a partnership. At the heart of the partnership is an agreement to provide integrated products and services. The first integration product will be released at the end of March.

According to Stacey Schneider, Hyperic’s senior director of marketing, the partnership is natural fit for both companies. “We were both talking to one of the country’s largest healthcare providers who wanted integration between our top-level systems and application management tools along with OpenNMS’ network administration tools,” she said. “We’ve run across each other often at customer sites and it made sense to formalize a partnership.

The first integration phase will meld together various alerts and allow customers to navigate between Hyperic’s HQ product and The OpenNMS Group’s OpenNMS platform. Next week, the two companies will lay out their roadmap for future integrations.

What might this partnership mean in day-to-day terms for data centers? Take a large distributed data center with lots of network traffic, Web servers and database servers. Any problems experienced by end users require IT operations personnel to access information about all aspects of the stack. OpenNMS can monitor the traffic network while Hyperic HQ can monitor application-level performance metrics such as CPU consumption.

The idea, said Schneider, is to give IT operations a more complete view of their environments so that they can get at the root of problems.

San Diego, Calif.-based Verari Systems has joined Sun Microsystems and Rackable Systems, Inc. in the data center mobilization trend with its new Verari FOREST containerized data center.

These mobile data center offerings address the need for space and low cost power. Companies without room for more servers can buy one of these containers for less than it would cost to build a brick and mortar data center, fill it with servers and storage, and plop it down in an area where electricity costs are low - like an abandoned coal mine in Japan.

Case in point, Verari’s FOREST container can house up to 1400 blade-based compute servers or nearly 12 petabytes of blade-based storage using Verari’s new BladeRack 2 X-Series platforms in the modular unit.

Sun’s containerized data center, project BlackBox, is in a standard metal shipping container — 20 feet long, eight feet wide and eight feet tall — and can house about 250 single unit rack servers. Introduced in March 2007, Rackable Systems Inc.’s Concentro is a 40 foot by 8 foot shipping container that houses up to 1,200 of Rackable Systems’ rack-mount DC powered servers and up to 3.5 petabytes of storage. The company came out with a denser mobile data center called ICE cube in October. IBM came out with one, the Scalable Modular Data Center in July 2007.

Who first invented the idea of a mobile data center? Probably Google, which patented the containerized data center in October, and which reportedly has been using them for its own purposes since long before Sun or anyone else did. Whether the patent will cause issues for other vendors has yet to be seen, as no infringement suits have been filed yet.

But for all this activity, mobile data centers aren’t exactly selling like hotcakes; Sun announced its first customer in June 2007, and Rackable hasn’t disclosed any, other than saying they have some.

Should you use a drop ceiling in your data center design? It shrinks the amount of room you have for equipment, but it can also provide a return air plenum to help isolate hot and cold air. There is an interesting comment string on IT Knowledge Exchange, talking about the pros and cons of each approach.

Our data center design expert Bob McFarlane recommends a minimum data center ceiling height of 14 feet. But as you go higher, gas fire suppression systems become increasingly expensive.

If you do go with a drop ceiling, McFarlane recommends perforated metal pan tiles with high-density fiberglass insulation behind them, rather than standard office ceiling tiles.

What is your take on drop ceilings in the data center?

Please check out this site if you feel like laughing. It’s called, simply, SOA Facts. Its purpose is to make fun of service-oriented architecture, commonly called SOA, which is usually defined as building a set of business services that integrate data and processes from otherwise disconnected applications. Honestly, that’s the simplest definition I could come up with.

Anyway, SOA Facts nails the oft-ridiculousness of vendors reciting it as the end-all, be-all by making outlandish statements about the technology. A couple samples:

• SOA thought Mensa too easy, so it founded Sensa. SOA is the only member.
• Kazakstahn uses SOA to produce the world’s best Potassium.
• SOA is the only TLA (Three Letter Acronym) you will ever need. Until you actually implement it - then you’ll also need DOA.

If you’ve got some extra time afterward (we know you do), check out its sister site, Chuck Norris Facts, which includes this gem:”Chuck Norris destroyed the periodic table, because he only recognizes the element of surprise.”

This is pretty impressive; colocation provider 365 Main has been taking advantage of Pacific Gas & Electic Corp.’s (PG&E) power saving program and saved $54,000 in utility costs at its San Francisco data center in 2007.

In all, 365 Main saved 7,477 kWh in 2007 by curtailing power usage below its energy-consumption baseline.

How did they do it? The first step 365 Main takes is inviting their utility provider, PG&E, into their data center at the start of each year for an energy audit to find areas where efficiencies can be added, said Miles Kelly, VP of corporate strategy at 365 Main.

“For example, this year we expanded the areas that utilize motion controlled lighting in the data center and also reinsulated the condenser water lines for the buildings massive air conditioning system. We also reduce lighting levels in public areas during peak days,” Kelly said.

The data center host also adjusted their weekly and monthly generator test schedules to peak days, which reduces their utility consumption by about 8% during test periods. “This adjustment of the testing schedule allows us to reduce utility consumption without ever affecting the reliability or N+1 redundancy of our power system,” Kelly said.

PG&E has gotten a lot of attention this past year for its part in promoting power savings in data centers. Their Critical Peak Pricing (CPP) program encourages data centers to cut back power consumption by offering seasonal discounts to customers that reduce or shift their energy usage away from peak periods.

In addition to savings incentives, PG&E also works with companies like IBM to consolidate servers, reduce heat and cooling requirements in data centers.

And 365 Main has been pretty aggressive in the “green” data center effort as well, if not for the public relations benefits alone. Earlier this year, 365 Main promised to build all its future data centers in line withLEED certification, which is basically a nice label the U.S. Green Building Council gives to companies that make an effort to be environmentally responsible. The standard is said to be tough for power hungry data center to meet though, and is most commonly achieved in buildings with lots of office space.

There’s been tons of clamour from environmental interest groups warning data center managers that if steps aren’t taken to reduce consumption, power availability problems will continue to grow and costs will rise, and several vendors and industry organizations have developed online tools to measure data center efficiency and promote power savings.

One of the many tools is from West Kingston, R.I.-based American Power Conversion (APC), a power and cooling services company, which has a free Web-based tool that rates data center power efficiency.

So, come on people. Let’s save some power.