June 01, 2011
Cluster Technology Limited, better known as ClusterTech, was founded in 2000 to bring grid, cluster, and high performance computing technologies to research and enterprise. More recently they have expanded their focus to include cloud computing via their partnerships with GigaSpaces and Joyent to bring cloud services to China.
In their eleven years they’ve worked with a number of businesses and universities, including the Chinese University of Hong Kong where they helped establish the Center for Large-Scale Computation (CLC), which now provides financial engineering services to Asia’s prominent financial institutions. ClusterTech has not always been focused on high performance computing; it wasn’t until 2003 that they formally entered the HPC market in mainland China but since that time they’ve worked with China’s Earthquake Administration, ING, Bank of China, the China Meteorology Administration and the Chinese Academy of Sciences.
We recently asked the one of the company’s founders, CEO, and standing director of the CLC, Wai-Mo Suen about their involvement with high performance computing in China. Given their recent focus on cloud computing in Asia, we wanted to get a sense of the challenges that come with providing virtualized services in China.
On a side note, the company has a significant focus on research and development. The three founders, including Dr. Suen, are figures in academic physics. Dr. Suen obtained his Ph.D. from Caltech and is still a professor of theoretical physics at the Washington University. He is also an honorary professor of physics at the Chinese University of Hong Kong, where he taught physics alongside his two other co-founders.
HPCc: Describe your roots in high performance computing; what leaps did you need to make to go from physical HPC to some of your new cloud initiatives.
Suen: We started in year 2000 with the vision that many companies in China could benefit from advanced computing technologies. We developed a two prone approach: On the software side, we developed HPC software, e.g., management software for HPC clusters, middle-ware that enabled single thread codes to conveniently take advantage of clusters, and job distribution software for grid computing. On the computational science side, we provide consultancy services to companies to help them develop algorithms and applications on HPC platforms.
In the past few years, we realized more and more that, even on the software side, what many companies in the region needed most is in fact also consultancy and professional service….”For my problem what technology gives me the best solution, and how can I implement it?” With this in mind, we developed professional service teams to help companies in China to tap into advanced computing technologies, and to help vendors of advanced technology worldwide to enter the China market. Joyent and GigaSpaces are two examples in the cloud direction, among other directions.
We are particularly interested in cloud, as in our view cloud technologies can and will bring HPC capability to benefit the general public directly. It has the potential of facilitating an important paradigm shift for HPC: Traditionally HPC architecture focuses on solving large scale problems that are tightly coupled. While cloud technologies will help make such usage more resource efficient and energy efficient, in our view it is more significant that, with the rise of web and mobile applications, and the booming of industries like on-line games, social networking and E-commerce, next generation data centers need advanced cloud technologies to make efficient use of their hardware resources to meet the demands of the general public users.
HPCc: What do you provide for users of HPC in the cloud computing context. Note that your own general description is VERY general and vague: “Cluster, Grid, Cloud and other high performance computing (HPC) technologies to improve operational efficiency, address large-scale computing demand, develop Business Intelligence, and ensure high availability.” – Go beyond this.
Suen: For example, we have helped major banks in China and Asia Pacific setup HPC clusters/private clouds for: (i) To provide computing and market data feed services to the trading floor, compliance and risk control departments, as well as back office. These services reduced the time to crunch pricing and simulation algorithms, improved availability, and lowered total cost of ownership. (ii) To increase the capability while lowering the cost in handling processes involving large data base, e.g., the generation of monthly statements for credit card customers. And, (iii) To serve as a trading platform and platform for back-testing foreign exchange and other algorithms. Examples in other sectors include imaging in oil exploration in the Oil & Gas sector, and weather forecast and pollutant simulations in the public sector.
HPCc: Explain the details of your partnership with Joyent—what HPC cloud services will you be providing; what’s your end and Joyent’s end?
Suen: As the preferred professional service provider of Joyent in Greater China, we provide professional services including consultation, implementation, training, and level 1 and 2 support to data centers using Joyent technologies, Joyent OEM partners, and service providers based on Joyent. Together with Joyent, we have setup cloud services in Beijing and Hong Kong, which will expand to other cities in China, and join the Global Compute Network based on Joyent technology.
HPCc: What are some of the challenges for doing cloud business in China versus elsewhere?
Suen: Public policy and IT cost structure are the two key challenges. One needs to notify the government before publishing any publicly assessable commercial content on the internet in China. Therefore, we cannot let the end users buy and setup smart machines on their own. Cloud service providers are required to apply for government approval for each end user, resulting in higher administrative cost and a more complex cloud service workflow. Furthermore, cost structure of internet services is drastically different from that in the U.S. and other parts of the world. For example, cost of bandwidth is significantly higher but the market price for a virtual machine is lower than that in the US, which means a tighter margin for cloud operators. This demands a highly efficient cloud platform. This is one important reason for our choosing Joyent as the key partner in providing cloud service in China.
HPCc: How do you overcome arguments about the challenges of HPC clouds—especially in the public cloud sense—when there are serious data movement issues that make it a difficult fit.
Suen: Not all HPC applications are data intensive. Some are compute intensive and with high demand for bursting. Further, bandwidth is getting cheaper and technologies are emerging to make terabyte size file transfer more reliable and faster by the days. On the cloud backend, the connection between nodes are improving rapidly, and high performance file systems to support the needs for massive data movement before the start of applications are being developed. With emerging better programming paradigms like MapReduce for massive data processing in cloud platforms, cloud software are rack-aware, and become more and more intelligent to move computation to data, instead of moving massive data around. Lastly, as opposed to common practice in public cloud that stores data in a centralize network storage space, we have introduced a distributed local storage architecture in our cloud environment. Data can be stored locally to the compute unit minimizing data movement overhead in high performance cloud computing.
HPCc: What are some changes or new offerings, case studies or other items we can expect to see in the coming year from your company?
Suen: In the cloud direction, we are working on moving mobile (gaming and online business) and machine to machine (M2M) applications onto our own and our clients’ cloud platforms. We believe these are the fastest growing segments. These applications benefit from highly scalable and low latency cloud computing environments. As such, we are going to offer a Node.js platform as a service to encourage local developers to start making new generation of multi-platform (smart phone, tablet, PC, netbook) and multi-user games and applications.
May 16, 2013 |
When it comes to cloud, long distances mean unacceptably high latencies. Researchers from the University of Bonn in Germany examined those latency issues of doing CFD modeling in the cloud by utilizing a common CFD and its utilization in HPC instance types including both CPU and GPU cores of Amazon EC2.
May 10, 2013 |
Australian visual effects company, Animal Logic, is considering a move to the public cloud.
May 10, 2013 |
Program provides cash awards up to $10,000 for the best open-source end-user applications deployed on 100G network.
May 08, 2013 |
For engineers looking to leverage high-performance computing, the accessibility of a cloud-based approach is a powerful draw, but there are costs that may not be readily apparent.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/02/2012 | AMD | Developers today are just beginning to explore the potential of heterogeneous computing, but the potential for this new paradigm is huge. This brief article reviews how the technology might impact a range of application development areas, including client experiences and cloud-based data management. As platforms like OpenCL continue to evolve, the benefits of heterogeneous computing will become even more accessible. Use this quick article to jump-start your own thinking on heterogeneous computing.