These decision support applications typically comprise at their core discrete optimisation algorithms and large data warehouses.
The business benefits to be gained from putting high performance computing to use in these application domains are huge in principle. How high performance computing can be successfully put to use, however, depends on an intricate interplay between commercial conditions, additional benefits to be specifically gained from parallelism, hardware and software technology as well as software engineering efficiency. We assess present technology trends on the basis of our experience at Lufthansa Systems over the last few years and our plans for the time to come and we formulate our expectations for technology development in the near future.
We report in particular on work done at Lufthansa Systems in the areas of scheduling, budget planning and market data analysis as well as on work in progress in parallel crew rostering.
Hardware platforms used were HP-Convex SPP-1200 besides K-series servers while we prepare for a transition to V-class now. Software development is based purely on C++ with a strong trend towards use of base class libraries.