Tooling Up to Go Green - Systems Management News On The Web

DEPARTMENTS

HOME

TOP STORIES

DATA CENTER NEWS

ON THE WEB

PRINT EDITION

BZ MEDIA

AS OF 1/7/2009 9:12AM EST

Tooling Up to Go Green
By Jeff Feinman

June 16, 2008 — The name of the data center game these days is green, as in energy efficiency and reduced expenses.

Much of the IT industry is putting its focus on decreasing data center power consumption, mainly because of customer demand, and that it can result in a significant drop in operations cost. There are various ways in which this is being done.

Some organizations are seeking to institute standards for data center management. The Green Grid, for instance, has created two metrics, Power Usage Effectiveness and Data Center Efficiency, which help data center operators estimate the energy efficiency of their data centers. Also, Leadership in Energy and Environmental Design is a third-party certification program that provides data center ratings.

However, there is no clear data center standard, and there is difficulty in defining exactly what a green data center should entail due to different sizes and needs of facilities.

While data center standards may not be quite ready for prime time, many companies are already implementing power-saving, cost-efficient technologies, both as products on the market and for their own operations. Whether it’s phase-changing batteries, technologies that remotely power down servers or server consolidation efforts, many companies are trying their best to bring energy-reducing technologies to the table.

Everything Green
IBM has committed to a number of goals in reducing energy consumption with Project Big Green. About a year into the project, IBM has worked with more than 2,000 customers worldwide building data centers and creating energy-efficiency initiatives. The company is also partaking in a large server consolidation effort that it hopes to complete by the end of 2009.

Project Big Green attempts to conserve energy in every possible aspect of data center management, ranging from the facilities level to the administrator level. The project is aiming at implementing more intelligent design and more efficient cooling systems (for example, using liquid as opposed to air for cooling). It is trying to increase virtualization and put into practice what the company calls active energy management, which involves managing and monitoring data centers for optimal energy use.

There has been a push in the industry as a whole to “go green” and adopt data center management methods that cut energy costs and employ environment-friendly processes, so IBM is not alone in that respect. It has, however, rolled out some notable products within Project Big Green.

IBM executives said that between 55 percent and 65 percent of the power coming into data centers is used by HVAC systems, lighting and other maintenance factors, so IT equipment is using less than half of data center power. To try and shift those figures to a more efficient level, IBM has created the cold battery, which aims to improve the efficiency of data center chillers.

The cold battery is a phase-state changing material that can turn from a liquid into a solid. It is a better option for chillers than water, IBM said, because the temperature at which it turns to a solid can be set, rather than solidifying at the freezing point. This can use less energy when charging the battery. The cold battery can be used as a primary chiller or as a means of disaster recovery during power outages.

The cold battery is part of a cooling tower, a source for cooling the air in data centers. Executives said the battery holds power better than a traditional cooler and it can be charged at night when it is colder and utility rates are lower.

Project Big Green also brings to light IBM’s Active Energy Manager, which is included in all of the company’s servers and storage devices. Active Energy Manager allows users to manipulate the amount of power being used when a server is running. It can run on Windows on System x and Linux on each of System p, System x and System z.

At the system level, IBM has created the Rear Door Heat eXchanger, which is a water-cooled exchanger attached to the back of servers that removes heat generated in the data center.

“Water is far more efficient at dissipating heat than air is, and it can take as much as 60 percent of heat out right at the source,” said Rich Lechner, IBM’s vice president of enterprise systems.

At some point in 2008, IBM will introduce a new version of the Rear Door Heat eXchanger, which Lechner said will remove as much as 100 percent of heat generated from servers.

Other energy reduction efforts from IBM include thermal sensors for fans and products in the company’s Tivoli IT management line. Sensors inside of servers help fans spin as fast as is necessary to cool off the server. The Tivoli Usage and Accounting Manager can allocate costs of energy usage by department or user and track the use of resources.

Down the road, Lechner said IBM will release products to address data growth and application efficiency, along with more advancements to its liquid-cooling methods. Additionally, the company will continue with its recycling efforts to help assuage growing e-waste problems. IBM claimed to have become the first company to recycle 1 billion pounds of IT equipment last fall, and Lechner said the company will carry on recycling 40,000 systems per week and design equipment to be more recyclable.

Project Big Green is not just a way for IBM to promote energy-efficient practices around the technology industry, but also a means to improve its own methods. IBM has set a goal to double its server capacity by 2010 without increasing energy consumption. IBM has more than 10,000 servers around the world, and it identified 3,900 of them for server consolidation. Those will be consolidated onto 33 mainframes running Linux. Lechner estimated this will save more than US$25 million in energy use.

“Once we do that, we can dramatically scale the capacity of those mainframes without significant increase in power consumption, as opposed to the distributed world where your capacity is almost linear in relation to your increase in power consumption,” Lechner said.

A Meeting of Minds
Many major companies today have a green initiative similar to that of Project Big Green. “I think there’s a lot of pressure from customers being concerned with how green a company is and whether or not they want to do business with that company,” said Winston Bumpus, a director for the Green Grid. “It’s an interesting time where we have energy challenges and environmental challenges and there seems to be huge support for being as energy efficient as possible.

The Green Grid was founded in 2007 as a global consortium to promote energy efficiency in data centers and business computing. Some of its members include AMD, Dell, HP, IBM, Intel, Microsoft, Sun and VMware.

The Green Grid said that to save data center power, people should identify the servers that are using alarming amounts of power and enable server processor power-saving features, like the ability to remotely turn off servers when not in use. New technologies like Systems Management Architecture for Server Hardware (SMASH) can remotely power down servers into a sleep state. SMASH, developed by the Distributed Management Task Force (DMTF), is a Web services-based suite of specifications with architectural semantics and industry standard protocols for data center management. Back in February, the Green Grid formed a relationship with the DMTF to make sure that all information on power and processor utilization is easy to gather, Bumpus said.

It is also important for organizations to develop an air management strategy for cooling, according to the Green Grid. A few simple tips from the Green Grid to conserve energy include moving cooling closer to the servers, maintaining higher operating temperatures, and utilizing dynamic controls and higher specification.

“A lot of cooling levels that people have in place today are based on equipment of a generation or so ago, and a lot of the equipment today can run at higher temperatures. Doing that can save quite a bit of cooling energy,” Bumpus said.

Some companies say the flow of air can be crucial in effective cooling. BlackMesh, a Herndon, Va.-based managed technology company, alternates the back of its data center racks so the hot sides of the rack get equal cooling, which the company said is a 40 percent decrease in the electricity being drawn to cool the equipment. Jason Ford, CTO of BlackMesh, compared this method to baseboard heating of rooms with a circular rotation, allowing the flow of hot air to return to the system and the cold air to be sucked into the server.

Consolidate and Virtualize
Companies can concentrate servers in blades or racks to increase memory and the number of processors, or they can implement virtualization. “The more you’re able to consolidate and virtualize your infrastructure, there will be a concurrent reduction in size of the footprint of the data center,” said Jeffrey Hill, a senior research analyst with the Aberdeen Group.

“Similarly, with storage, you can use de-duplication techniques or tiered storage with rotating data that doesn’t need to be accessed away from online storage. The net effect is that virtualization and consolidation allow people to reduce data center floor space, which means I’m reducing aggregate cooling requirements and power consumption.”

Virtualization is playing a large role in data centers, as organizations are reducing the number of servers at 5-to-1 and even 10-to-1 ratios. This can increase the amount of productivity on a per-server basis, and organizations like the Green Grid are trying to implement best practices for server consolidation and virtualization in data centers.

Novell is a company whose data center approach consists of server consolidation, hardware migration and data relocation. These three factors exist mainly within products and technologies acquired from PlateSpin, a company that creates virtual machine (VM) and server consolidation products.

Converting physical machines VMs and the ability to remove a service from the hardware required is Novell’s focus with its PowerConvert workload portability product. The product, which Novell gained in the February acquisition of PlateSpin, streamlines data center workloads between physical servers, blade infrastructures, virtual hosts and backup archives.

Novell has also flaunted its Forge disaster recovery and data restore product, also a former PlateSpin offering. “Because it’s constantly taking a snapshot of your existing servers, the product has a complete backup of that environment if for some reason the data center were to experience a hardware failure or a complete meltdown,” said Richard Whitehead, director of product marketing for Novell’s systems and resource management group.

Whitehead said that Novell also has initiatives to ensure that the company’s code within all of its products is as efficient as possible. The more efficiently that code is written to carry out tasks that don’t require a lot of processing power, the less work the machine will have to do, thus reducing power, he said.

Novell’s SuperLab testing facility in Utah tries to conserve power by viewing several machines through a single monitor, which Whitehead said is probably typical in most data centers. Also, the SuperLab runs virtual environments so multiple services can run on a single server.

One data center company that has delved into virtualization technology is IT company Tideway Systems. Tideway has a topology system that maps business applications to an underlying physical and virtual infrastructure and finds servers that have no dependencies on them. Tideway’s Foundation application dependency mapping product does an IP scan of subnets to find IP-facing devices, according to the company. For the server hardware, the product then runs commands to view the network communications and configuration files to see dependencies.

“We use quite sophisticated pattern-matching techniques to work out what we’re seeing,” said Kosten Metreweli, vice president of product marketing for Tideway. “Obviously, in a data center, things are deployed in very different ways. You need to be flexible for data center activities.”

Metreweli said in addition to this server consolidation and decommissioning capability, Tideway can also cut power for organizations with virtualization initiatives. The company uses VMware technology and can consolidate physical servers on an 8-to-1 ratio. Metreweli claimed some Tideway customers have been able to identify 30 percent of servers to virtualize, and that translated into saving $1.8 million per year, along with 968 tons of carbon per year.

Data Center Snapshot
Avocent, a Huntsville, Ala. IT infrastructure management product provider, has created software solutions to help data center managers view capacity and consumption. The company’s DSView 3 is data center management software that monitors all connected IT and network devices, the company said. Data center functionality is put into a single interface and allows administrators to remotely access devices on multiple platforms at numerous locations.

“We will capture the power consumption information from the power distribution units and then aggregate that information at all different levels within the data center,” said Ashish Moondra, a product manager at Avocent.

“What I mean by different levels is you can aggregate that for an entire rack or row of racks. The reason we’re doing that is that once you have that information down to that granular level, it becomes a tool that makes sense for a lot of different people working in the data center.”

DSView 3 can help systems administrators, IT managers and CIOs obtain information about their organization’s data centers, Moondra said. It offers a snapshot of the data center power levels, and the organization can create comparative reports between specific racks or power distribution units.

Force10 Networks brings a high-density system with a backplane that can work with many ports. Executives of the San Jose-based network builder said this is done by implementing a passive data center model, which lowers costs and doesn’t use active components. By using less devices, there is less consumption and less heat, which reduces HVAC needs.

“One of the bottlenecks is how to build a backplane that can scale and support a lot of ports cost effectively,” said Stephen Garrison, vice president of marketing for Force10 Networks.

“Think of a giant PC board in the back of a switch acting as a motherboard connecting components. That is still a very difficult problem. Some people do it by using optical technology or active components to help push the signals through the copper and not have issues. So we had to invent some new ways to do that on a passive model.”

Garrison said that some companies talk about three-layer structures, with an access layer, an aggregation layer and a core layer. This is a way to get around not having high-density systems, as an organization will have to build horizontally to scale without high density. With high-density systems, the company said, it is possible to scale by having more ports per server.

“We don’t preach three layers,” Garrison said. “We look at the cabling infrastructure. Sometimes customers want a top-of-rack solution that aggregates back to a core versus a direct connect solution with no intermediate devices.”

Whether it be IBM’s server consolidation efforts or the Green Grid’s creation of metrics, it’s obvious that organizations want more energy efficiency to reduce costs and be more environment-friendly. With companies big and small working to provide more energy-efficient data center products, everyone seems to be on board the ship going green.

Related Search Term(s): Cooling, data centers, green computing, HVAC, power, virtualization, Avocent, BlackMesh, Force10 Networks, IBM, Novell, Tideway

Share this link: http://www.sysmannews.com/link/32382

EMAIL THIS ARTICLE

SEND FEEDBACK

MORE SPECIAL REPORTS