EGI TRANSITION WOKSHOP

SUMMARY - EGI TRANSITION WORKSHOP

The EGI Transition Workshop was organised during the EGEE’08 conference in Istanbul (Turkey) on the 22nd of September, 2008. The workshop focused on the transition from present EU project based e-Infrastructures to a long term sustainable EGI model. The EGI Transition Workshop was followed by a closed EGI Policy Board Meeting.

First Session

The workshop organiser Per Öster (CSC) welcomed more than 100 participants to the workshop and presented the programme of the day. The workshop was organized in two sessions; the first part included presentations held by representatives from EGI design study project and selected e-infrastructure projects. The second part was a panel discussion.

Ludek Matyska, EGI_DS Project Director, held a presentation “EGI Blueprint or Do not be afraid of EGI”. He stressed that the European Grid Initiative is just a culmination of national grid efforts, and there is no reason to fear the EGI. Matyska gave a brief look on the EGI Blueprint draft that Dieter Kranzlmüller had presented more in detail earlier during the Opening Plenarium. Matyska focused on the construction of EGI, NGIs, implications on national and international levels, and the transition to the EGI model. Matyska stated that the EGI organization (EGI.org) is a small coordination body governed by National Grid Initiatives (NGIs) and through them by users and resource providers.

Bob Jones, Project Director of the EGEE-III project, scoped on the transition from EGEE-III to EGI from the EGEE’s point of view. He listed recommendations made by EGEE for the EGI Blueprint, and presented a transition plan. The transition plan details the steps needed to migrate EGEE’s operations to EGI. It identifies major risks and shortcomings, and develops stop gap strategies in case that a full transition seems unlikely in the remaining lifetime of EGEE-III. The transition plan may result in a modified Description of Work for EGEE-III updated for the second year.

Ognjen Prnjat from SEE-GRID-SCI (SEE-GRID e-Infrastructure for regional eScience) made a presentation about the South-East European initiatives for regional e-Infrastructure sustainability. SEE-GRID-SCI aims to engage international user communities e.g. meteorology, seismology, environmental protection its work. It tries to provide application-specific service extensions and infrastructure for new communities, to consolidate actions towards long-term sustainability and European Grid Initiative inclusion, and to strengthen the regional and national human networks. One of the core objectives is to develop and strengthen the coordination and cooperation of national e-Infrastructure programmes in the region of South-East Europe.

Prnjat presented a path towards sustainability from the SEE-GRID-SCI’s perspective:

- SEEREN1/2: establishing the regional inter-NRN interconnectivity and GEANT links

- SEEGRID1/2: building the regional Grid infrastructure within and beyond EGEE

-  SEEFIRE: studying the feasibility of long-term solutions for dark fiber backbone in the region

-   SEELIGHT: implementation of the lambda facility in the region

-  BSI: Caucasus region connections

- SEE-GRID-SCI: e-Infrastructure for large-scale environmental science: meteorology, seismology, env. protection. Inclusion of Caucasus.

-  SEERA-EI: regional programme managers collaboration towards common e-Infrastructure vision and strategy

EGI is seen as a viable model, however with some features that do not cater for regional sustainability and collaboration. He underlined a need for a more dynamic, free and flexible model for scientific collaborations. SEEGRID would like to see the EGI.org only as a purely operational coordination and harmonization body.

Hermann Lederer from DEISA 2 project held a presentation about DEISA2 User Communities, HPC ecosystem and EGI. DEISA2 project aims to consolidate the existing DEISA infrastructure and to continue activities and services that currently contribute to the effective support of world-class computational science in <place w:st="on">Europe</place>. A focus is on the provisioning and operating the specific infrastructure services which allow its users to efficiently work within a distributed high performance computing environment.

DEISA2 will try to evolve the DEISA infrastructure towards a robust and persistent European HPC ecosystem, by enhancing the existing services, including support for European Virtual Communities, collaborating with new European initiatives; especially with PRACE that will enable shared European PetaFlop/s supercomputer systems. It will also advance the existing distributed European HPC environment to a turnkey operational solution for a persistent European HPC infrastructure. According to Lederer, it is expected that the concept for an integrated HPC infrastructure will essentially be based on national supercomputing centres of the member states. Conceptually this is very similar to the EGI approach with the NGI as the basis. Therefore there is a high need for cooperation of EGI and DEISA in the areas of infrastructure services and user support. Key areas include the user community support in the various scientific disciplines, common AAAA (Authentication, Authorization, Accounting and Auditing), an access to data repositories and high speed networks.

Second Session

The panel discussion had the following representatives:
- DEISA2: Hermann Lederer
- D4Science: Pagano Pasquale
- PARADE: Peter Kunszt
- SEE-GRID-SCI: Ognjen Prnjat
- EELA-2: Bernard Marechal
- WLCG: Jamie Shiers
- DORII: Marcin Plociennik
- BalticGrid2: Algimantas Juozapavicius
Per Öster (CSC / EGI_DS) acted as a moderator of the panel.

Usage Drivers
An access to the most powerful supercomputing architectures is needed in Europe. It is vital to collaborate and to share data and resources, not just to obtain resources. There is a need for dynamic, new groups building collaborations and a new dynamic ecosystem setting up of VOs.

Major challenges in operating grid The support for single projects is sufficient, but the support for VOs in unstructured communities is a new and large challenge. Issues of autonomy and competition have to be dealt, as well as the peer review and the technical HPC conformity review process. The challenge is to offer a stable service while architecture, middleware, applications are changing.

What EGI needs to provide?
- There is a need to foster different scientific communities to collaborate.
- The current EU funding scheme is project based.  In the EGI, the numbers are too fuzzy and there is nothing concrete to present to the government. EGI follows a very traditional infrastructure model (OS, VO etc...).
- The role of VOs has to be more important. Additionally, there need to be better tools to maintain them and to enforce policy decisions.
- NGIs cannot establish international projects; the role of the EGI is to do that.
- Biggest requirement is the continuity in the current high level of service.