Air-conditioning Equipment Control through AI
Continuing from FY 2016, we worked to improve air-conditioning energy efficiency inside and outside Japan. We are improving efficiency through new technologies to achieve Environmental Action Plan (Stage VIII) targets.
For example, we reduced annual power consumption by 20% by using Fujitsu's innovative just-in-time modeling air-conditioning controls, predicting temperatures and humidity an hour ahead from temperature, humidity, and power data for outside air environment and inside servers, and then controlling outside air cooling and air-conditioning equipment air temperature. Moreover, we plan to further expand the control area. We are also currently verifying AI-driven learning control technology. We are improving power efficiency for air conditioning by modeling optimal air conditioning properties based on factors including wind volume, processing heat, and ICT device load. Moving forward, we will utilize the effectiveness of these measures to enable control of cold/heat-source facilities such as chillers and cold-water pumps.
Improved Operation through Development of a Data Center Evaluation Tool
At the Fujitsu data center located in Tochigi Prefecture, as a method of evaluating whether or not the cooling energy of ICT devices is being efficiently operated and controlled, we conducted analysis based on the heat balance and airflow balance. During analysis, we ascertained the energy reduction potential amount from the current structure and facilities specifications of the data center. We then developed a "data center evaluation tool" capable of calculating the theoretical PUE value which can be achieved through improvements.
Using the tool makes it possible to analyze the necessity of implementing high-efficiency facilities by visualizing energy loss and proposing optimal related measures. Based on the analysis results, we are working to clarify policies for improving data center operation and achieving even greater efficiency. Also, in order to realize even smoother improvements, we are planning to create guidelines based on expertise gained from evaluation methods and different data centers. We will then make these guidelines available throughout the Fujitsu Group.
Expanded Use of Renewable Energy
In May 2017, we announced our company's Environmental Vision for the achievement of a low-carbon society. We are taking steps to systematically transition to renewable energy for the power used by our data centers. In particular, the shift to renewable energy is proceeding with a focus on overseas, where it is possible to procure green energy. Renewable energy now accounts for 17% of the total power used at our 36 main data centers inside and outside Japan.
Almost all of our data centers using renewable energy are located overseas. We will work to actively expand the use of renewable energy from our overseas sites.
Data Center Goal of "Extreme Energy Conservation: Working Toward PUE 1.0"
Expansion of high-performance computing (high-performance, high-heat-generating servers used for big data and AI) increases energy used to cool data centers. Accordingly, Fujitsu participated in the Ministry of the Environment's FY 2016 Low Carbon Technology Research, Development and Demonstration Program, and is working on fundamental reduction of CO2 at data centers.
We are focusing on cooling methods suited to ICT device heating levels, working particularly on high-heat-generating servers. Using a liquid-immersion cooling method in which ICT devices are submerged in fluorinated inert liquid (Fluorinert) with high thermal transport efficiency and insulating properties, we are challenging the "PUE 1.00 wall" with Fujitsu's new proprietary natural convection technology for Fluorinert circulation. AI-driven improvement of cooling efficiency is also being verified.
Establishment of a Laboratory for Low-carbon Next-generation Cloud Infrastructure
In April 2017, Osaka University, Fujitsu Limited, and Fujitsu Laboratories Ltd. established the "Next Generation Cloud Research Alliance Laboratories" in Osaka University's Cybermedia Center.
The Laboratory will be used for research on energy-saving technology and security technology toward development of a next-generation cloud infrastructure that will be compatible with the future low-carbon society. In addition, it will be used to train the next generation of technicians, mainly for Osaka University students. In the future, the three organizations will conduct proof-of-concept testing based on research findings, at the Laboratory and on test beds (verification platform) built internally and externally by consortiums in which the Laboratory participates. Further, the developed energy-saving technologies will be utilized in an effort to make the next-generation cloud infrastructure and next-generation AI infrastructure more energy-conserving and low-carbon. Fujitsu will strive to apply the results of development in its AI technology "Zinrai."
Cold Storage Geo Replication Technology
Currently, coordination between data centers is conducted via two methods: low-delay short-range synchronous communication and long-range asynchronous communication backup. With the former, it is highly possible that multiple data centers would suffer damage in a large-scale disaster, and with the latter, there is increased data delay, so it wa only possible to use a standby redundant connections.
On the other hand, "cold data," mainly images and video, which is rarely updated, constitutes the majority of data. This means that there is an increasing need to accumulate new data and share data between multiple sites. However, issues hindering data center coordination between multiple sites include synchronization between data centers and increasing data reading response speed. In that context, we have been working since November 2016 to conduct joint verification testing on intercontinental data center coordination, focusing on cold data storage, between a data center inside Osaka University and a data center of Paris-based AntemetA. We have built a long-range coordination storage infrastructure, which, in addition to improving throughput and disaster resilience, excels at lowering costs by eliminating redundancy within individual data centers, and are verifying the effects.
Contributions to Industry Organizations
Concentrating a customer's ICT assets in a datacenter also contributes as an energy-saving benefit for society as a whole. By actively being involved in working group activities in the various industry organizations related to datacenters, Fujitsu helps to enhance the value of datacenters to society. Citing one example, in particular, Fujitsu is leading efforts to devise and promote the use of PUE (Power Usage Effectiveness) measurement and calculation methods as an industry representative to the Japan Data Center Council (JDCC).
Visualization of Effects
Fujitsu has won approval from the Ministry of Economy, Trade and Industry's J-Credit Scheme (for reduction of CO2 emissions in Japan) for a method for calculating CO2 reductions achieved by switching to a Fujitsu datacenter and for a scheme that will actually issue credits. This makes it possible to visualize a customer's environmental contribution from outsourcing datacenter services to Fujitsu and to credit the customer for that contribution.