GRID Large Panel Making real large-scale grids for real

GRID@Large Panel “Making real large-scale grids for real money-making users: why, how and when? ” Achim Streit 4 August 2005 Forschungszentrum Jülich in der Helmholtz-Gesellschaft

Large? 4. . . in number of participating sites 4 virtual organization management 4. . . in compute capability 4 resource management 4. . . in storage capacity 4 data and replica management, data access and transfer 4. . . in geographical expansion 4 network latency, bandwidth, availability 4. . . in amount of users 4 authentication/authorization, accounting/billing 4. . . in middleware platforms 4 interoperability 2 Forschungszentrum Jülich

Important aspects in large-scale Grids 4 easy and straightforward installation and configuration of Grid middleware platform 4 full interoperability of Grid middleware platforms, if more than one is used 4 seamless integration of new sites and resources in the Grid infrastructure 4 scalability and redundancy of major components and services for e. g. information, scheduling, accounting, . . . 4 smooth and fast integration of new users at all sites 4 common rules, policies, and methods for e. g. security mechanisms, directory & file naming conventions, . . . 4 powerful monitoring system for jobs and resources 3 Forschungszentrum Jülich

DEISA’s approach 4 UNICORE as Grid middleware platform 4 DEISA primer for new site administrators with all information about naming conventions, operational methods, configuration files, hostnames and ports of core services, . . . 4 dedicated network from NRENs and GEANT between the DEISA sites for job migration and data transfer 4 GPFS as storage and access to data 4 LDAP for user data and integration in UNICORE User Data. Base (UUDB) 4 LL-MC and LSF for resource management, co-allocation, job re-routing, and migration 4 Ganglia, Inca, and MDS 2 for automated testing and monitoring 4 Forschungszentrum Jülich