Gregor Molan is currently head of Science and Research at Comtrade. Since 1997 he has merged software engineering with research on algorithms based on theory from formal languages to graph theory and type theory. Slovenian and EU research grants finance research projects. Decades of experience in software development and system architecture, combined with ample research experience, form a unique background for his current role as a research lead, primarily focused on storage with artificial intelligence and its applications.
Windows Client on the Working CFFS Environment Based on CERN EOS
- Stopnja 400
-
Datum
torek
27. september 2022 12:00
Comtrade 360 has been involved in the development of CERN EOS since 2015. This cooperation has recently produced a Comdrate 360 product: the CFFS (Comtrade Fast File System). EOS-wnc, a native client for the Windows operating system, is an essential element of this system. An example of a working CFFS environment for an end customer will be shown. Experiments at CERN have strict requirements for high bandwidth, low latency, high reliability, and high availability for the collected data and subsequent analysis, which the EOS Open Storage platform has successfully fulfilled since 2011. EOS is mainly built and running on a Linux-based operating system and does not provide a native client for Windows. Despite the current trend of moving towards cloud data storage services and cloud execution of processes, the problems and weaknesses brought by remote cloud services are increasingly being revealed. This opens up new formal requirements for direct influence on the determination of the level of reliability, the level of availability, the level of security of stored data, and, above all, legal requirements for the ownership of the physical data. Although the EOS performance has been demonstrated by the IT Storage and Data Management group at CERN for the LHC experiments, this is not enough for customers to trust their data storage to CFFS. A concrete working CFFS environment is required and is currently being set up in Ljubljana (on a smaller scale) to work as a high-performance data cluster, identical to the CERN cluster used to store experiment data. Identical hardware for Linux and Windows clients with high-performance connections to the CFFS cluster proves that the most demanding data access and processing requirements for experiments at CERN can be met even on Windows. The measurements show the required data transfer speed and low latency on the Windows operating system. This presentation aims to demonstrate CFFS in action and the transfer of CERN storage to Windows clients.