KEN Brill: Engine Crisis in data center

The 2nd Annual "2012 China Data Center Industry Development Conference" was held in Beijing on the 20,112-month 17th, uptime&http://www.aliyun.com/zixun/aggregation/37954.html ">nbsp;i Nstitute founder Ken Brill made a video speech at the meeting. Ken Brill shared uptime Cato's four main areas of research, Ken Brill said Moore's law was a problem of excessive energy consumption, although each data center alone would increase energy efficiency but not keep up with the cost. Ken Brill found that the energy consumption of IT equipment purchased at $10,000 per year was at least doubled over the past decade. This means that the energy consumption will be twice times the same, or more, on the premise that the IT hardware is unchanged. The problem may not be apparent in China, as data centers do not run long in China, but they are inevitable in the future.

Representative: Hello, because Mr. Ken Brill is too busy to attend the meeting in person. Below I provide some Chinese translation, I start now, please put the video. Thank you for the moderator's introduction, Today I and you discuss the integration of data infrastructure, first of all, I briefly introduce myself, I regret that today I can not personally attend the meeting in Beijing, have to use the modern network technology has been exchanged, I have visited China 8 times in the past 2 years, I hope that we can introduce this technology into China, I am fortunate enough to have a lot of Chinese industry colleagues, but also to the development of China's current data center to understand. It is a pity that I cannot attend the meeting here today and thank you again, and I hope you will understand.

More than 30 years ago, I was in the data center power supply, 21 years ago I created the uptime Cato, I in the data center engineering design has put forward some innovative ideas, now has become the industry standard, I have developed the data Center classification standard, the other little known is that I collect, organize data center downtime, As well as data center operating costs, I have done a lot of research and several books have been published, and I received a lifetime achievement award from the Datacenter Conference in 08, followed by a graduate of Harvard Business School.

Here's a description of what I'm going to do today, and I'll start with the work we've done in uptime Cato, and then focus on our four major research areas to help us better understand the future of the data center.

Uptime Cato uphold the concept of from the user, to the user, our members pooled computer computing users, we hope to ensure the reliability of the data center operation. 25 years ago, people were not very familiar with data center failure technology, but this situation has changed, we are independent of third-party suppliers advisory body. Uptime Institute1993 and North America, has now expanded to the global, 02 into Europe, this year into the Asia-Pacific region, and hope to come to China in 2013, the above is the Uptime Cato overall introduction.

Here is a talk about the specific technical work carried out by uptime Cato, which shows a large data center of something, we can see the data center as an industry, power through the diagram of the switch control into the data center, the control room is small space, once the control room failure, Affect the reliability of the data center. The grading system of Uptime Cato is proposed for this purpose. The two operators in the figure are maintaining the switchgear to ensure the reliability of the power equipment, and they are overhauling the circuit, which is extremely high in voltage and can reach 125,000 volts.

We can see one of the staff members preparing to leave, another in front of the protective masks and gloves, the following photos were taken by closed-circuit cameras, at this time all normal, followed by an explosion. Although I was not aware of the injury, it was clear that the staff member had been injured and the data center had stopped working. The picture is dark and the equipment has stopped working.

Show you just a few photos because this is closely related to the work of Uptime Cato, our goal is to ensure the normal operation of the data center, Uptime Cato in this PPT listed in the three aspects of the action, the first by the Uptime Cato Enterprise User Association, China Merchants Bank and Huawei are our founding members in the Asia-Pacific region, the founding member conference will be held in June this year in Singapore, we hope to attract more Chinese enterprises. Uptime Cato now has more than 110 member companies worldwide, these are large enterprises, UPS load more than 40 billion, members will share and learn, and sharing best practices. The purpose of the association is to serve its members.

In addition, we provide certification services for enterprises other than members, we are divided into four levels, most enterprises want to achieve level three or four levels of standards, of which three of the hope that safe maintenance does not affect IT operations. In fact, there is no stage in the data center, the number of research errors. In China people may not know the grading standard very well, we provide the standard at the same time also provide training, to help the data center to apply the grading standards, the classification standard is very concise, the standard set the definition of each level and the corresponding implementation of the goal, for the number of generators and the amount of power supply and other specific design plans do not provide. A very common problem is that many data centers claim to be able to reach level three or fourth level standards, but not in practice. We did a survey, they may only reach the second level, or the first level. Therefore, we have certified the data center, respectively, for the topology and the actual operation. Both are very important, we hope that more and more data centers in China can be certified, in addition to our professional certification. Third, we also organize public education. The 2012 Uptime Institut Congress will be held on May 14, which is what we call public education.

In the next few slides I will share with you the questions we raised during the seminar, I raised the question of the high energy consumption caused by Moore's Law 6 and 7 years ago, Moore is one of the founders of Intel Moore's Law, which is that in the next 5 years, the number of integrated circuits will increase by one times every 24 months, The IT revolution that we are seeing today is bound to continue. But what I'm talking about is the high cost. Each generation of IT equipment energy efficiency problems, can not pursue our data requirements. Although each data center alone, energy efficiency increases, but not to keep up with the increase in costs this is what I call the Moore's law costs too much of the problem.

The problem has sparked concern in a few years, as the engine of the data center has been at risk. We found that the energy consumption of IT equipment purchased at $10,000 per year was at least doubled over the end of the past decade. This means that the energy consumption will be twice times the same, or more, on the premise that the IT hardware is unchanged. This is what we call the data center engine crisis, a problem that may not be apparent in China because the data center is not running long in China, but it is inevitable in the future.

Let's look at this slide, the change in the efficiency of the data center. These concepts are mutually reinforcing, and we work with McKinsey, a world-renowned consulting management firm, to discover that the cost of equipment is not integrated, as is the case in China. If we can synthesize the above factors, it is not difficult to find that the equipment accounts for 8% of it expenditure, the cost of equipment at a rate of 20% per year growth. The reason for this is the high cost of Moore's law that we mentioned before, and the increase in it energy consumption will affect everything else. Many people are concerned about the ever-increasing cost of electricity, which actually costs more than a hundredfold to manage spending, 10 times or 20 times times as much as the server itself. The increase in costs will lead to basic spending and increased energy consumption.

In the United States and China, it and equipment belong to different departments under the jurisdiction of the lack of corresponding communication, change the development model, we must link these two aspects. This has proven to be a 50% reduction in energy consumption on the basis of less additional capital and equipment input, which is what we call a change in the efficiency of data centers.

Finally, I would talk about the integration of the data infrastructure, a phenomenon that is rapidly emerging, namely that it equipment is merged in an innovative way, and that its driving force is also the problem of the high cost of Moore's law. So now there are cloud computing technologies for public and private clouds, software that serves, infrastructure services, and so on, to reduce 8% of equipment in operating costs. The integration of different equipment comes from the increase of cost. My data shows that the average server is only 2% of the time, but consumes the energy needed for continuous operation. As a result, digital infrastructure integration can improve usage and optimize performance without increasing input.

Before closing, let me go back to the topic of the presentation, the integration of data infrastructure. I would like to invite you to participate in the seminar we will be organizing on May 14, a total of 130 theme chapters, you are welcome to send your message through the mail, we provide training for the most qualified companies, and finally thank you for your attendance, and hope that the next opportunity to be able to communicate face-to-face, thank you.

(Responsible editor: The good of the Legacy)