Case Studies - Private Sector
Many of OCF’s customers would prefer not to be identified by name as they regard their research infrastructure as confidential. For this reason some of the case studies below are only identified general industrial area. Should anyone wish to discuss such systems further we will attempt to promote a private dialogue.
Customer : Landmark Information Group
Requirements : Landmark needed a storage environment which pools storage hardware, to cope better with current storage demands
Solution : OCF designed a vendor-independent, scalable, flexible, reliable, virtualized storage environment
More Information...
Customer: The National Nuclear Laboratory (NNL)
Requirements: The NNL uses its skills, expertise and facilities to provide technical support to customers such as Sellafield Ltd and the Atomic Weapons Establishment. It also plays a key role in the nuclear industry by helping to reduce clean-up costs, maintaining critical skills, and attracting new people to the industry.
Solution: Replacing a Silicon Graphics based HPC system; in 2007 NNL launched ‘Gemstone’ (its unique name was the winning submission from a staff competition). Gemstone’s design incorporates a powerful cluster of 56 IBM System x3550 Servers with Intel Woodcrest processors and 10 IBM System x3455 Servers with AMD Opteron processors. A combination of Intel and AMD processors enables NNL to take advantage of the fact that some codes for producing 4D models operate more effectively on different processors.
NNL’s new HPC system creates and integrates up to 7 different 4D computer based models to produce a single set of results for customers – this enables its customers to undertake safer, more efficient and cost effective nuclear projects. Gemstone makes a single integrated set of results available in just 5 days, a decrease from up to 4 weeks if 4D models are processed in series.
Testimonial : Read more…
Customer: Tecnica
Requirements: Tecnica UK Ltd is an automotive engineering design consultancy. Tecnica offers a hot forging process design service. Tecnica's engineers are capable of creating 3D computer simulations of the forging process for its customers.
Solution: Live from February 2008, OCF designed, implemented, configured and now supports a bespoke HPC system, which is providing significant improvements in the power and performance available to process 3D simulations for Tecnica.
OCF’s HPC system design – based on a cluster of servers - consists of both an IBM System x3655 Server with AMD Opteron Dual-Core processors - with 4GB of memory and six 146 GB SAS DISKS - and IBM System x3455 Servers with AMD Opteron Dual-Core processors – with 8GB of memory.
The new HPC system is reducing processing time of Tecnica’s extremely complex 3D simulations by up to 80 per cent (40 hours down to 8 hours).
Testimonial: Read more…
Requirements : A highly available Linux cluster to run CFD and FEA ISV codes together with a SAN configured in a parallel file system and tape back up. This was the company’s first significant investment in cluster technology and hence it was vital that the system was easy to manage and quality training provided.
Solution : A 512 core Linux Cluster utilizing IBM System x servers with Intel Woodcrest processors. A percentage of the nodes were configured with large amounts of memory to facilitate large FEA models and the head nodes configured in a failover configuration. 24tb of SATA disk SAN storage was provided via an IBM DS4700 controller configured under IBM GPFS, and an IBM TS3310 tape library for back up utilizing IBM Tivoli software. The cluster was designed with dual networks-a gigabit Ethernet management network and a Voltaire Infiniband high performance network facilitating extremely fast compute and storage I/O. IBM CSM was used to manage the cluster and the scheduling is provided by PBS Pro from Altair. OCF have dealt with this customer for a number of years and have supplied large numbers of workstations to run CAD applications, firstly from SGI and latterly fro HP and IBM. We have also supplied IBM UNIX servers in the past and a number of IBM SAN storage systems.
Customer : Grand Prix Racing Team
Requirements : A large, highly available, exceptionally fast Linux cluster to run CFD applications, primarily Fluent, together with a large SAN storage facility configured as a parallel file system and a new datacenter.
Solution : OCF, together with its partner IBM, designed and implemented a new datacenter with full redundancy and room for significant system expansion. The cluster was originally configured with 512 AMD Opteron cores in IBM System 1u servers in 2006 managed via IBM’s CSM, with two networks-a gigabit Ethernet management network and a high speed Infiniband network from Cisco. 100tb of SAN storage was provided with a combination of FC and SATA disc in IBM DS4800 storage systems configured as a parallel file system using IBM GPFS with an IBM TS3500 tape library and Tivoli software for back up. The cluster system was doubled in size in 2007 by the addition of a further 552 AMD Opteron cores again housed in IBM System x 1u servers.
OCF have dealt with this customer for a number of years and we have also supplied smaller cluster systems, again utilizing IBM 1u server technology, a large cluster of graphics workstations utilizing 16 IBM System x3755 quad processor servers with AMD Opteron processors, each with 128gb of memory and nVidia FX5600 graphics cards and the existing Infiniband network. We have supplied a large number of IBM workstations for the CAD applications an, following IBM’s withdrawal of the desktop workstation product we are transitioning the customer to Lenovo workstations with great success.
Customer : A Tier 1 Luxury Motor Car Manufacturer
Requirements : A fully upgraded compute cluster environment to run FEA and CFD applications.
Solution : OCF and its partner IBM designed and implemented two distinct environments for these demanding applications. For the CFD cluster we supplied a 780 core Linux cluster utilizing IBM System x 1u servers and AMD Opteron dual core processors with two networks-a gigabit Ethernet management network and a Voltaire Infiniband high performance network. For the FEA cluster we supplied a cluster of 4 IBM p575 8 processor servers running AIX. OCF has dealt with this customer for a number of years and in June 2006 supplied it with a 322 core cluster utilizing IBM e326m servers with AMD Opteron processors and a full Myrinet high speed interconnect.
Customer : Prestige Sports Car Manufacturer
Requirements : This customer required a fast, scaleable, easily managed, reliable Linux Compute cluster to run CFD applications, with SAN storage under a parallel file system.
Solution : OCF designed and implemented a 100 core Linux cluster utilizing IBM Bladecenter H technology with two networks-a gigabit Ethernet management network and a Voltaire Infiniband high performance network. The SAN was configured as an IBM DS4700 with 16tb of SATA storage utilizing IBM GPFS parallel file system. Some of the nodes were configured as high memory for particular applications requiring large amounts of memory and the system was managed with IBM CSM. To facilitate exceptional throughput and efficiency we supplied the MOAB scheduling software from Cluster Resources.
Customer : A Global Aircraft Manufacturer
Requirement : A high speed, reliable Linux Compute cluster to run both ISV and in house CFD code.
Solution : OCF designed and implemented a 524 core system in IBM Bladecenter systems utilizing AMD Opteron processors with a dual network-a gigabit Ethernet management network and a high performance Myrinet fast interconnect. The system was managed with IBM CSM.
Customer : Laing O’Rourke
Requirements : A High Performance Visualisation Collaboration Centre to enable project teams to work collaboratively and to enable customers and potential customers to experience accurate simulated environments.
Solution : OCF, with its partners Fakespace Systems and Orad Hi-Tech Systems, designed and implemented a fully immersive system with two display environments-a large solid screen display and a smaller mobile display, with Christie Mirage projectors set up in active stereo mode and a full room sound and vision control system. The graphics image generation systems from Orad provided the fastest possible image generation at the time. The project was complicated by the fact that the original location for the system, Heathrow Terminal 5, became unavailable and it was necessary for OCF and its partners to re-engineer the solution to fit a new build facility at the customers Dartford head office. The company has credited the facility with being central to its success in winning a lot of work for the 2012 Olympics.
Customer : A Major Global Pharmaceutical Company
Requirements : A High Performance Storage solution to support cross-referencing of millions of compounds against a constantly increasing flow of targets.
Solution : an IBM DS4000 series storage environment utilising GPFS General Parallel Filesystem to deliver extreme throughput for data reads & writes
More Information...
