Gentzsch, WolfgangKowalczyk, Ryszard2019-05-062019-05-062007978-3-88579-211-6https://dl.gi.de/handle/20.500.12116/22146Nowadays, proprietary and monolithic IT environments are more and more transformed into standard- and component-based service infrastructures. In this keynote, we will analyse this trend especially for the area of e-Science. While in the past decades, scientist have developed centralised, application-focused (silo-oriented) hardware and software environments, they now (for good reasons) tend to join forces in building and operating general-purpose, networked, distributed compute and data grids, recently called e-Infrastructures. This trend is enabled by grand technology progress towards reusable, decomposable, service-oriented software architectures and standard off-the-shelf hardware, among others. However, it is still challenging to design, build, and operate such complex infrastructures for e-Science. Therefore, we have analysed several major (grid) infrastructure projects around the world (such as TeraGrid in the US, NAREGI in Japan, the e-Science Initiative in the UK, and DEISA and EGEE in Europe) for the sole purpose of extracting their lessons learned and valuable recommendations to be followed while designing, building, and operating our own German D-Grid infrastructure and user and provider communities. This keynote will present these findings combined with our own lessons learned. A catalogue of recommendations for those interesting in building similar e-infrastructures will conclude this presentation.enText/Conference Paper1617-5468