ISC HPC Blog
The Different Colors of HPC
Wolfgang Gentzsch, Chair of the ISC Cloud Conference series and President of the UberCloud Community and Computing Marketplace talks about the pros and cons of HPC on workstations, on in-house servers, and in the cloud, and invites you to the Rolls-Royce Panel on Monday at ISC.
There is no doubt that the benefits especially for small and medium size enterprises (SMEs) of using HPC within their design and development processes can be huge. For example, producing better quality products, enjoying a higher return on investment (ROI), avoiding product failure early in the design process, and shortening time to market. Why then are most of the engineers and scientists still running simulations just on their workstations, when many are regularly dissatisfied with the performance? The main reason is that the alternatives present some significant challenges.
The first alternative, buying your own HPC server, comes with high total cost of ownership (TCO) as has been demonstrated by IDC. In addition to the cost for the server, expenses for staffing, training, software, downtime, and maintenance can easily account for ten times the server cost over three years. In addition, there are often long internal procurement and approval processes. And for many, ROI is not clear, although it is expected to be huge according to a recent study by IDC.
The second alternative for SMEs is cloud computing, which allows users to experience the benefits from high performance computing without having to buy and operate their own HPC system. HPC in the cloud (or HPC as a service) allows engineers and scientists to continue using their own desktop systems for daily design and development, and to submit (burst) the larger, more complex, time-consuming jobs into the cloud. Benefits of the HPC cloud solution (in addition to HPC in general) include, among others, on-demand access to ‘infinite’ resources, pay per use, reduced capital expenditure (CAPEX), greater business agility, and being able to dynamically scale resources up and down as needed. In short: better, faster, cheaper!
However, HPC in the Cloud comes with challenges too. For one thing, It presents a new business and working paradigm, for the manager as well as for the engineer, while at the same time, security, privacy, and trust in service providers can become significant issues. Also, conservative software licensing is only slowly including the pay-per-use service model. Finally, Internet bandwidth is often not able to accommodate the heavy data transfer needs of a cloud computing.
The UberCloud HPC Experiment was developed to deal with those myriad challenges. It provides a platform for engineers and researchers to discover, explore, and understand the end-to-end process of accessing and using HPC Clouds, and to identify and resolve the major hurdles. Since July 2012, this experiment has attracted over 1200 organizations from 72 countries, leading to the creation of 152 teams in CFD, FEM, life sciences, and big data. SMEs who want to develop better products faster are invited to join the free UberCloud HPC Experiment to explore HPC as a service, in the cloud.
At ISC in Leipzig, on Monday June 23 at 2:00 PM, jet engines will take off in the cloud. We have invited speakers from Rolls-Royce Germany and from cloud provider CPU 24/7 to discuss their real-life UberCloud project of running a complex aero-thermal analysis of a jet engine high-pressure compressor assembly in the cloud. The session will highlight the benefits, challenges, and lessons learned.