The modular HP PODs were delivered to Airbus sites in Toulouse and Hamburg, Germany. Each POD contains all the elements of an HP Converged Infrastructure, including servers, storage, networking, software, management, and integrated power and cooling. A total of 2,016 clustered HP ProLiant BL280 G6 blade servers enable the two 12 meter-long containers to deliver the equivalent of nearly 1,000 square meters of data center space.
The HP PODs have enabled Airbus to quickly expand data center capacity, boosting computing performance for aircraft development while saving space and energy. Compared to an installation in a nearby customer data center, the water-cooled HP PODs consume up to 40 percent less power. With a near-optimum Power Usage Effectiveness (PUE) rating of 1.25 from The Green Grid(TM) consortium,(1) Airbus decreased operating expenses while delivering power capacity in excess of 15 KW/m(2).
HP Enterprise Services provides data center services for Airbus, while HP Technology Services managed the transportation, deployment and commissioning at the customer site. For this project, HP managed the preparation of the new infrastructure, the technology decisions and the technical elements of the transition.
Airbus is clearly serious about HPC. I poked around on the Internet and came across a brief about a presentation Guus Dekkers, CIO of EADS and Airbus, will do at the International Supercomputing Conference to be held in Hamburg in June 2012. (EADS, by the way, stands for European Aeronautic Defense and Space Company N.V., a global pan-European aerospace and defense company comprised of Airbus, Eurocopter, Astrium, and Cassidian.)
In the brief describing Dekkers presentation, HPC is referred to as the enabler for the virtualization of aircraft development – a process that takes billions in investment and years’ in time. The design process must be flawless, and the sooner digital modeling can be used, the better. The brief notes that “numerical simulation is one of the most important means to realize this objective,” and to make that happen, the aircraft is modeled in an early phase representing both physical flight behavior as well as the behavior of the different aircrafts structures such as flaps) and systems. The design teams then apply different improvement plans and “what if” studies to this model, which allows for a quick analysis of the consequences of any modification or optimization suggested.
Substantial HPC power is needed for the necessary real-time simulations, incredible amounts of detail and continually condensed timeframes in which to get the job done. To get a sense of just how much calculating and data creation occurs, the definition of the overall aircraft model ultimately translates into between 500 million to 1 billion nonlinear equations which have to be resolved during each reiteration in a time-accurate manner.