Application practice of cloud storage in Tsinghua University

Whether in the industry or in academia, cloud computing is undoubtedly a big hot spot. As the highest institution of higher learning in China, the Institute of High Performance Computing technology of Tsinghua University has been in the leading position in the field of cloud computing theory research and application. Tsinghua University has pioneered many attempts and practices in high-performance computing and cloud computing, and has been widely used in schools. In this respect, IT168 reporter interviewed Tsinghua University computer Science and Technology Department of High-performance Computing Institute Professor Wu Yongwei, to explore Tsinghua University in the cloud storage and high-performance computing areas of construction and application practice.

"Each of us produces a lot of data every day, but there is no time or energy, or there is no good platform for us to manage the data." For example, I will take DV photos for my family, will be a variety of data, but if the CD is broken, there is no way. With the development of the information age, everyone produces more and more data, but everyone's management of the data is more and more weak, so I think the primary goal of cloud storage is to provide everyone with management data, storage and backup conditions, and do not lose, because now the data has become a legacy. "I believe that most people share the views shared by Professor Wu Yongwei, and it is based on this point that he is optimistic about the application and development of cloud storage."

Professor Wu Yongwei, his research work mainly covers two aspects, on the one hand, cloud storage, which has also spent the most energy in this regard, emphasizes the usefulness of cloud storage: "We tend to make cloud storage more practical, experimental environments to test the advancement of our technology and the availability of the system." "On the other hand, virtual computing, similar to Amazon's elastic Computing Cloud (EC2), a virtual cluster built by the computer department of Tsinghua University can be used to provide high performance computing courses in schools, providing virtual computing environments and operating environments in the form of browsers or Web clients, To meet the needs of teaching experiment in high performance computing.

Shared activation mass Storage

Tsinghua University's Corsair cloud storage service is designed for the staff in Tsinghua University and college students in the mass storage warehouse, the user registered to obtain large capacity of private storage space, through the Corsair client easy access to a variety of learning, software, audio and video and games and other information, in addition, Tsinghua University has introduced the concept of community design into mass storage, "in the cloud storage, the amount of data generated will become more and more, the same data to create a common interest of people will be aggregated into a collective, we call the community." "Users can create communities based on their hobbies, and serve as community administrators, simply describe the community and then share it, and all users will see the community, interested students can apply to join the community, through this approach to build the community network." Therefore, in the cloud storage built in Tsinghua University, in addition to the traditional data backup, there are community data sharing, such as a number of well-known teaching video and courseware in public storage space for the school students to use.

▲corsair Cloud storage Services

The sharing of data makes the data stored by more students use, play the maximum value of the data, at the same time, data sharing also inspired a new application. Professor Wu Yongwei, based on such a large amount of data, can provide many application services. For example, there is a popular video program on the campus of Tsinghua University, "Kangxi is Coming", Tsinghua University has a wide range of wireless network, so through mobile phones and WiFi, students can watch the program, the program video content from the campus of the cloud storage.

It is understood that at present Tsinghua University Campus cloud storage scale has reached 100TB, deployed in different geographical location of three storage nodes.

Build a solid platform to open up

At present, Tsinghua University is limited to the use of cloud storage in schools, while at seven or eight universities and some software parks in the country, and not open to society. Professor Wu Yongwei that the main reason is that cloud storage is based on data as the core of the application, so the requirements of the network is relatively high, the campus network provides a good platform and network environment, for the use of cloud storage provides favorable conditions to create a good user experience. In addition, it is also due to the consideration of data security and sensitivity.

Speaking of future development plans, Professor Wu Yongwei said: "We want to link the cloud storage in different regions, such as Shanghai University students, may join a Tsinghua community, so that people can produce a broader sense of data sharing." ”

Professor Wu Yongwei introduces cloud storage as a base platform, and then develops more applications and services on the data based infrastructure, just like the small apps that are now ubiquitous on mobile phones. But the premise is to do the basic platform well, otherwise the application will not be discussed. "and the platform technology content is higher than the application technology, from the university point of view, we pay more attention to do the system structure, the platform well." When the platform is really done, we encourage openness and provide a platform interface for everyone to develop their own applications based on our platform, like Apple or domestic Baidu. We will certainly be in this direction in the future, but what we have to do now is to do the platform well. "he said.

Cloud Storage Underlying architecture

In the context of the underlying architecture of cloud storage, Tsinghua University adopts open source Distributed file system, and on this basis, for personal storage did some optimization and improvement, he introduced: "For ordinary users of the file storage system will be relatively large amount of data, so we designed a distributed metadata management system;" Another example is that personal document files in many cases need to be modified, so how to improve the user experience is more important, we spent more time and energy in these areas. ”

In addition, Professor Wu Yongwei revealed that in Tsinghua University, the cloud storage experimental platform, has begun to use self-developed Distributed File System.

For distributed processing, the amount of data facing is usually large, also need a lot of hard disk, and through the collaboration of software and hardware to achieve stability, reliability. As a non-commercial research unit, Tsinghua University's storage platform is also available free of charge to the campus, so in terms of performance and cost balance, Tsinghua University has its own considerations. "Our cloud storage applications use a lot of Dell storage devices, and we're also going to expand the device by buying a lot of hard drives." "said Professor Wu Yongwei.