2 page Case Study - Posted 6/12/2008
Views: 417
Rate This Evidence:

Daresbury Laboratory

Facility for Breakthrough Science Seeks to Expand User Base with Windows-Based HPC

The Distributed Computing Group (DCG) at Daresbury Laboratory provides high-performance computing and advisory services to the academic community in the United Kingdom. For its newest cluster, the DCG chose a dual-boot system with Windows® HPC Server 2008 and now benefits from streamlined deployment and enhanced cluster management capabilities. The new cluster will expand user choice and increase ease of use with the familiar Windows environment.

Business Needs

Daresbury Laboratory is one of the main research facilities within the Science and Technology Facilities Council in the United Kingdom. Located in Warrington, Daresbury has 550 staff members and offers research facilities to the university consortium throughout the country. The 75-person Computational Science and Engineering (CSE) Department at Daresbury, of which the Distributed Computing Group (DCG) is a part, hosts and runs national high-performance computing (HPC) facilities as well as smaller systems for the purpose of evaluating new HPC technologies.

*
* Windows HPC Server 2008 will help us extend our user base by taking high-performance computing to new user communities in a way we were unable to do before.  *
Jonathan Follows
Group Leader, Distributed Computing Group
Daresbury Laboratory
*
CSE develops and provides powerful simulation codes. It places particular emphasis on the advancement of basic computation methods to tackle new scientific challenges on new generations of hardware. Core areas that are supported include quantum chemistry, molecular simulation, solid-state physics, materials simulation, engineering, and environmental simulations. The United Kingdom academic community uses CSE services to test breakthrough science theories. One such example is in the field of aircraft design with the creation and modeling of codes that measure the effects of turbulence on helicopter blades. CSE users number in the hundreds and benefit from expedited research with large-scale computing resources.

As a provider of the latest generation in high-performance computing services, the DCG wanted to continue to offer the latest technologies. It also sought to increase its user base beyond traditional HPC users and offer the academic community new computing choices.

Solution

The DCG decided to create a new cluster based on Windows® HPC Server 2008. Daresbury chose Windows HPC Server 2008 because of its strong management features, including the new heat map view of cluster utilization and the Job Scheduler capabilities based on service-oriented architecture. “The new functionalities offered by Windows HPC Server 2008 provide great potential,” says Igor Kozin, Computer Scientist at the Distributed Computing Group, Daresbury Laboratory.

The new cluster has 32 dual-socket compute nodes (HP ProLiant DL160 G5 server computers), quad-core Intel Xeon processors, and ConnectX DDR InfiniBand interconnect switches. The cluster includes two head nodes and is a dual-boot system with Windows HPC Server 2008 and SUSE Linux 10.1. One head node is configured with Windows whereas the other head node is configured with Linux; all the compute nodes have two internal disk drives and can be booted with either operating system.

Daresbury relied on its partner, Streamline Computing, to deliver the cluster with Linux already installed. In May 2008, the DCG deployed Windows HPC Server 2008 internally in a short time frame using Windows Deployment Services.

Benefits

Windows HPC Server 2008 provides an easy-to-deploy solution that makes it possible for the DCG to give users more options for their high-performance computing needs. Additionally, Windows HPC Server 2008 presents an easy-to-use interface with solid functionalities. Benefits include the following:

  • Streamlined deployment and management. Windows HPC Server 2008 has proven easy to deploy and manage with Windows Deployment Services and a simplified graphical user interface (GUI) and wizard. “When installing Windows HPC Server 2008, we were able to install all 32 nodes in one easy run,” says Kozin. Additionally, the department had applications installed by the end of the deployment week. “Porting codes from Linux to Windows HPC Server 2008 was very easy and painless. I was running the ported code within a day,” says Kozin. New functionalities offered by Windows HPC Server 2008 facilitate cluster management. “I like the Windows HPC Server 2008 heat map―it gives you the ability to see the load on your cluster across a variety of different parameters,” says Kozin. The team also benefited from increased insight. “With Windows HPC Server 2008, I could easily connect to compute nodes. I had a full GUI to every compute node and I could see exactly what was going on,” says Kozin.
  • Increased ease of use with solid functionalities. DCG employees benefit from the comprehensive GUI offered by Windows HPC Server 2008. “A strong point of Windows HPC Server 2008 is the very solid graphical user interface, which in many respects, Linux still lacks,” says Kozin. Users can also manage their project requests from start to finish using familiar applications. Therefore, they radically reduce the time required to become productive in the new environment. 
  • Expanded choices to increase user base. Daresbury looks to expand its user base and give the academic community more options. Windows HPC Server 2008 will make it possible for users who are less technologically advanced to access supercomputing facilities. “Windows HPC Server 2008 will help us extend our user base by taking high-performance computing to new user communities in a way we were unable to do before,” says Jonathan Follows, Group Leader, Distributed Computing Group at Daresbury Laboratory. Also, the addition of Windows HPC Server 2008 in a dual-boot environment expands actual operating system choices. “When customers ask us whether it is going to be a Linux or Windows system, we can now offer both options,” says Follows.
This case study is for informational purposes only. MICROSOFT MAKES NO WARRANTIES, EXPRESS OR IMPLIED, IN THIS SUMMARY.
Document published June 2008
Solution Overview



Organization Size: 550 employees

Organization Profile

One of the largest research facilities in the United Kingdom, Daresbury Laboratory provides high-performance computing services to the academic community throughout the country.


Hardware
  • ConnectX DDR InfiniBand interconnect switches
  • HP ProLiant DL160 G5 server computers
  • Intel Xeon processors

Software and Services
Windows HPC Server 2008

Vertical Industries
Public Research Centers

Country/Region
United Kingdom

Partner(s)
Streamline Computing