As compared to a year or two ago, many data center operators are now not keen on the word "green". This is not to say that they have abandoned projects designed to improve energy efficiency and make better use of IT resources, as they can both cut operating costs and reduce environmental impact.
After reviewing the six green practices that have been accepted by a large group of organizations and institutions to reduce the cost of data center operations, the 451 group of market consulting companies released a comprehensive report on energy-efficient data center practice. Some of these practices, such as the comprehensive planning of data centres, the search for alternatives to traditional refrigeration systems, may already be well known, and other practices, such as the introduction of DC power and prefabricated data centres, are rarely mentioned or considered unsustainable.
When planning to upgrade their facilities, data center operators should consider or reconsider these trends. Of course, not all of these practices are suitable for every organization, but if you don't look at it, you may never know that there are such practices.
Practice 1: Adopt a comprehensive approach to improve overall efficiency
Data center operators need to approach each project in isolation by combining integrated technologies and tools to comprehensively address efficiency issues. This is not really a groundbreaking concept: Green grid organization Since its inception in 2008 has been the "overall consideration" slogan. One of the biggest benefits of doing this is to promote synergy among departments, especially facilities and IT departments, to ensure that all participants are able to agree on upgrades to the data center. For example, the IT department ordered a high-end server cabinet for mission-critical projects, so he only had to find out if there was enough power or cooling to support the cabinet when it reached the data center.
This is not the end of collaboration within the Organization. In an effort to reduce energy waste, many business sectors are now becoming more open, and what was once the most secretive of the departments could be disclosed. The most famous example of this is http://www.aliyun.com/zixun/aggregation/1560.html ">facebook, which uses its open computing program, and its blueprint for efficient servers, cabinets and cooling equipment is open source."
Practice 2: Smarter cooling Systems
Data center operators are becoming aware that data center facilities are not a large freezer for meat storage, but need to be properly cooled. Although IT managers do not want the device to fail because of overheating. But if you spend 50 cents per dollar on it for refrigeration-which is the average pue ratio in many traditional data centers-it's almost certain you're burning money.
There are many ways to reduce the cost of cooling, some of which are not accepted by businesses for fear of destroying their precious hardware. This includes liquid cooling technology, which is thought to have a lot of maintenance problems and has a limited range of uses. But a liquid cooling method from the green Revolution Cooling Company uses a low-cost, non-polluting insulating liquid that is said to absorb 1200 times times as much air.
As for air cooling, this free cooling method is more popular than liquid cooling. The concept is simple: just build the data center in colder places. Then the use of air energy-saving devices to absorb outdoor air, exhaust hot air, so as to achieve the purpose of cooling. More and more companies are starting to accept this free cooling method. For example, Microsoft has expanded its original 303000-square-foot data center in Dublin to 112000 square feet, and the expansion is entirely based on air energy-saving devices for cooling.
Practice 3: Using 17815.html "> Renewable energy generation
A group of companies, including Microsoft, Google, Facebook, Yahoo and Belle Canada, have been impressed by the installation of renewable power generation equipment to reduce electricity bills. Solar is a popular option, and as solar panels fall, this approach is becoming increasingly popular. The 451 Group noted that solar energy is not the only option, for example, Fujitsu has made helium power an alternative. And some companies are exploring the use of cow dung to generate electricity. While on-site power generation requires some upfront investment, in normal circumstances, this investment returns quickly.
Practice 4: Using DC power supply
In a typical data center environment, voltage converters abound from outdoor power switches to servers. Every time the voltage is converted, some power is lost. The initial voltage at the power switch is 16000VAC (alternating current), which is then converted to 440VAC to 220VAC and converted to 110VAC before reaching the UPS feed to each cabinet. Each UPS converts incoming alternating current to DC and converts it back to alternating current. Each UPS then assigns the AC power to each server. As much as 50% to 70% of the power that goes into the data center is wasted on this shifting path. The DC power supply system can eliminate all these conversion costs and greatly reduce energy expenditure.
There are several reasons why companies are suspicious of switching to DC power. Some operators may not yet know that hardware vendors are already offering systems that can support the DC environment. There are also cost considerations, such as high voltage power systems that require large installation costs. But saving 50% or more of electricity in many years is a huge payoff, and it's not counting the space savings of data centers that have reduced the number of swing devices.
Practice 5: Make it devices more efficient
Virtualization has helped countless businesses consolidate a lot of hardware and reduce the number of servers. The 451 Group report said that many companies were deploying DCIM (data center infrastructure Management) software. These tools graphically display all the physical and logical assets of the data center, showing the layout of cabinets and data center floors and the thermal load of cabinets. With this software, data center managers can model any migrations, additions, or changes to minimize the impact of performance by creating complex best-choice scenarios before the change is implemented.
On the hardware side, Google and Facebook, for example, use their own specially designed servers rather than off-the-shelf products. With two companies using technologies that remove redundant components, Facebook has even designed a dual-core processor to be placed side-by-side rather than back and forth on the motherboard because it improves air flow. On the storage side, solid-state storage systems also have the potential to reduce energy waste. The 451 group said: "The emergence of solid-state storage disk may become a game-changing, although the adoption of the market has only just begun." ”
Practice 6: Using modular routing
There are many pitfalls in the design and construction of data centers: the first is to spend a lot of time (and waste time), because the entire organization's departments need to send representatives to participate in the discussion of the new data Center building design issues, often controversial. Then there is the construction process will be wasteful, much more than the actual needs, such as data center size, density or redundancy. There may also be a need to pay for the construction of electricity and additional infrastructure that cannot deliver any returns.
As a result, data center operators are increasingly enamored of modular building and scaling. This model limits the deployment of removable containers, also called modules, which are made up of standardized components such as it hardware, power supplies, and cooling equipment. Companies can add new modules at will, depending on the increase in computing requirements, because of the mobile and stereotyped design of the modules. These modules also provide the flexibility to freely assemble in free space. (Wave Compilation)
(Responsible editor: The good of the Legacy)