Subject: notes from AM discussion From: Florida Estrella Date: 2013-04-22 13:34 To: Alberto Di Meglio , Balazs Konya Discussion notes: Alberto: Main points to clarify by the end of the day (1) EGI vs MEDIA (2) Technical roadmap independent of technical funding. Projects are tools to implement the vision. Steven: Multi year vision on the middleware innnovation linked with those who will consume is useful. Who is providing the vision, deliver the vision, and engage with the user communities - may not be the people in the room. Mirco: EMI is an asset for Europe and for its user community that should not be lost. MEDIA is a place where the new vision will be put forward, in an inclusive way. EGI ---- what is the level of commitment from PTs what is the level of value for customers Globus Forum (?) based on Apache didnt work. Be realistic on what can be sustained. Is this an attempt to hold the community until the next funding? dCache -------- Some good statements Some statements that should not be made EMI collaboration and methodology have improved dcache Work on something that is excluded in MEDIA, outreach Little money from government/funding for outreach Common representation to customers No benefit in lightweight endpoints Interoperability is guaranteed EGI is already doing some of the things like roadmap, technical collaboration, requirements - this is double work to deal with MEDIA and deal with EGI for the same thing NO interest in ticket systems, release management, repositories etc CERN Data --------- like to have forum for synchronization eg roadmap sharing. emphasis on synchronization rather than coordination. place to trigger targeted actions eg synched work on particular topics ways to connect to new users eg events participation NOT interested from MEDIA: -top down coordination -prioritization and rfc management -technical representation -services such as release managment, repos, testbeds, trackers EMIR ------ will benefit a lot from MEDIA collaboration with other products is very challenging communication channel with other products, offered by EMI EMIR will miss the best practices like Q&A, release management, testbed EMIR will miss distribution channels eg repo EMIR will miss communication channels ARC -------- EMT mailing list will be missed ARC invested heavily into the ff EMI activities that should be/remain production after EMI : EMI-ES, GLUE2, EMIR, CANL, synchronization with ARGUS and XAML profiles, CAR, EPEL, datalib ARC doesnt like an additional overhead, additional management layers eg send the same report to EGI and MEDIA CESNET --------- Agree with CERN Data. Everything that constitutes lightweight communication is welcome. Everything that raises the question who is paying eg shared testbed should be discussed and be concerned about. UNICORE ------- UNICORE is a small team Many comments and problems that others have do not apply to UNICORE CANL-Java is the only sync of UNICORE and EMI roadmaps EMI-ES is an evolution but still needs to move to production UNICORE had a lot to teach on security, but wasnt taken up in EMI Expectations: keep links with ARC, dCache and some areas of gLite Further collaborations with EMIR Like open repository where people can upload their sw EGI has offered a repository UNICORE not interested in release coordination, trackers MEDIA should not market itself as a mw; UNICORE does not want to be represented by MEDIA; will talk directly to EGI, PRACE. Observer 70% full member 30% gLite-Security: UH and NIKHEF -------------- coordination whereby if they make a release, it will not break other components and vise versa. like the idea of coordinated release to ensure we do not break stuff have direct talks with communities like biomed, EGI STS work within WLCG ongoing, should continue like this INFN: VOMS, ARGUS, WMS, CREAM, STORM, WNODES -------------------------------------------- no strong coordination no release management collaborate on getting the best to users eg requirements gathering, outreach, preparing for innovation, new paradigms, new partners, new proposals and projects common developments eg CANL work should continue standards: how to use standards, profiling, harmonization do not want roadmaps and prioritization no heavy weight single representation with users requirements gathering : for example EGI TCB CSIC: MPI Start ---------------- dependent on the CEs, at least be aware of CE developments ie roadmap sharing is important EMI-ES is a nice work and should continue; open to collaborate if there will be a PT on that dont need big top down enforcement of coordination best practices can be recommended, but no need for enforcement no need for repos and testbeds CINECA: ---------------- Technical shared with UNICORE Interested with federated authentication and EMI STS A similar initiative exists in the world, and follow their lead? Similar initiative HPSS which has similar approach, a legacy in many sites Collective requirements talking to different communities with different infrastructures. This is a clear requirement that should be addressed by MEDIA eg data movement from EGI to EUDAT QOSCOSGRID: Poznan ------------------ Provide integrated system for job management deployed in PLGRID Interested in promoting products in Europe MEDIA is not only about collaboration but also release sw in the MEDIA repository Better understand EGI vs MEDIA There is MoU with EGI, checking if there is opportunity to work with MEDIA