100 likes | 246 Views
Summary of morning sessions from the rapporteurs. John Gordon Rolf Rumler. Contents. Goal of the Workshop (Jürgen Knobloch) Overview and Summary of the EGI Blueprint Proposal (Dieter Kranzlmüller) EGI Operations (Tiziana Ferrari) Items from discussion Summary. Goal of the Workshop.
E N D
Summary of morning sessions from the rapporteurs John Gordon Rolf Rumler
Contents • Goal of the Workshop (Jürgen Knobloch) • Overview and Summary of the EGI Blueprint Proposal (Dieter Kranzlmüller) • EGI Operations (Tiziana Ferrari) • Items from discussion • Summary www.eu-egi.eu
Goal of the Workshop • NGI need to be established, to do that governments and funding bodies need to be convinced of the importance of a e-infrastructure to European science. • Comments • Justification for funding agencies is difficult as EGI is for a minority of researchers. (Unidentified participant) • Individual competence of workshop participants not sufficient to judge all parts of the blueprint. Further discussion and consultation with funders is unlikely to happen over the summer vacation. This could derail the target of a final draft by September (Unidentified participant) • The vision shown in the blueprint is too narrow which might lead to its rejection if presented to funders in its present form. (Malcolm Atkinson) www.eu-egi.eu
Overview and Summary of the EGI Blueprint Proposal DK reminded us of the overall EGI-DS vision He reported that some NGIs don’t expect to operate their own infrastructures. They want someone to do it for them. Stressed that the e-infrastructure needs to be neutral and open to all communities Proposed funding 21M€ per annum. Membership fees 1M€, EC 20M€ EGI Geneva Workshop www.eu-egi.eu 4
Overview and Summary of the EGI Blueprint Proposal • Q1. Having to support multiple middleware stacks is felt to be unrealistic because of interoperability problems. (Maria Dimou) • Q2. Do we expect EC funding for EGI.org in addition to the 1M€ membership fees? • DK 1M€ only for management. 10/45 FTE. • Mirco – Important that NGIs ‘own’ EGI.org. Membership fees aim to achieve that. Need to recognise that there may not initially be enough NGIs to form critical mass of funding for EGI.org • Q3 As well as expectations on NGIs, NGIs need to know the expectations on EGI.org in order to convince their national funders.(D-Grid) www.eu-egi.eu
Q4 How are large user communities involved and how are efforts coordinated especially with the supercomputing infrastructure? (Kunszt PRACE) • NGIs have to find mechanisms to integrate user communities • From the EGI-DS project point of view, there are initial links to GEANT, PRACE and others which need to be formalized • Q5 Is EGI for science or all communities? Connect all major facilities, ESFRI, and data and information. Vision should be broader and enable things to happen rather than restrict to technical issues.How do international organisations fit into the EGI scheme? (Atkinson UK) • Answer: this problem arises also in the networking context, there are multiple solutions. • Q6 We must not lose site of the fact that we are doing this for Science, not for NGIs. How do applications provide input. The blueprint assumes all interactions and requirement capture is through NGIs. Also need to take input from the scientific community at management level.(Bird, CERN)
EGI Operations and Security Tiziana Ferrari described the tasks expected to be performed by NGIs and by EGI.org. There is tension between local freedom of NGIs and guaranteeing interoperation across borders for international VOs. Experience of EGEE encourages NGIs to cluster and share experiences, costs and responsibilities. EGI Geneva Workshop www.eu-egi.eu 7
EGI Operations and Security • Q1 Items like ticketing for network problems, management of relationship between EGI and resource providers are missing. Concerns about merging networking with performance monitoring because technical competence might become obsolete: a specific networking activity will be needed. (Xavier Jeannin) Answer: find out what can be “sold” to NGIs, what developments and SLAs will be needed. • Q2 What is the position of ASGC in EGI. No answer yet. • Comment - Show that EGI isn’t just a continuation of EGEE. (Malcolm Atkinson) • Q3 Who does the technical assessment of requirements and prioritisation in the long run? (Malcolm Atkinson) Answer: under discussion, equivalent of TMB of EGEE would be a possibility. www.eu-egi.eu
Q4 Who trains the staff in systems administration and operations? (Malcolm Atkinson) Answer: not explicitly mentioned, should be addressed. • Q5 What is the justification for the central accounting repository? What is the operational model? Answer The justification is that it is not yet feasible for a VO to query the accounting repositories of all NGI in order to find their usage across EGI. The operational model is for NGIs to collect their own accounting data and to have a central EGI cache NGI needs this service for its own community which is larger than just the international VOs. They may also account on more variables than EGI. It is not the place of a lightwieght EGi.org to provide a service for local NGI use. • Q6 Are there any extra costs, like for supporting LHC/LCG for example? Answer: no extra costs expected.
Summary • Participants obviously not convinced that the blueprint in its current state meets all the requirements to convince national funding bodies to support it. • In particular, • Doesn’t address the whole of the ERA in particular international facilities on ESFRI Roadmap • Does not describe a mechanism to take input from the scientific community at board level • Have missed a few operational functions (networking, training) • Not obviously neutral and inclusive (DEISA, PRACE) www.eu-egi.eu