23rd June 2010 Special Topic: NGS Technical Roadmap - Notes by Pete Gronbech
NGS surgery. [1]
New services, new funding?
BDII for information services Computational Data Information SRB , iRODS needs better advertising. NGS people say SRB too heavy weight Authentication and Authorization Been running this service for a long time and very high level of service. SARoNGS service now available. Will split up NGS VO to provide finer grained choice for the users. Accounting. We must provide a uniform single accounting solution. Currently 3 solutions. RUS, APEL, GridSAFE (in HPC) Need a single solution to sit under NGS UAS to be able to show value to funding agencies. Move to utilise APEL and GridSAFE where appropriate. Not only computational grid so other (storage and network) need to be accounted. Monitoring. Need a single solution. Integrate INCA and Nagios. Need to develop reporters for all the services. User Access Interfaces: UI/WMS, portals HERMES,MEG, AHE etc, Need to make certain new communities integrate with teh NGS. Access to Compute Must be simple to deploy and maintain Need to be able to support new user communities. Engage with GridPP to install gLite CE at new sites, must acknowledge that we are not experts at this and should
EU IGE Integrating Globus in Europe. Can piggy back on their support structures. Investigate UNICORE and ARC are possible long term middleware providers. Eventually, a best of bread solution.
Grid SAM solution here in Oxford used by BADC Need to make certain they are not cutting off these communities un necessarily.
Access to Data Need to continue support for SRB and on to iRODS as required. LFC clients for AFS cells Interactive usage/advanced reservation. Continue the development of HARC. Applying for funding to support reservation of storage and network services. Summer internships in OERC.
Future Service: Visualization: VizNet Clouds: Want to move it into operations. JISC flexible services just been funded.
Conclusion. Large scale projects with diverse requirements. One size does not fit all. EGI independent from middle ware provider. Align UK technical roadmap with EGI technical roadmap.
Bham meeting lead by NGS for JISC sfree projects...
AR: Not just UK aligning with EGI NGS has been actively driving the plan. Neil Geddes and AR on executive boards. EGI project EGInspire. UK in group with Ireland and Netherlands. First 6 months AR is the rep. Then rotates. Staffing only just getting put in place. Will be similar to EGEE.
Grid PP side of NGI, continues to be business as usual. Need to continue to make smarter use of the joined up setup. Should not develop two solutions. Question about the setup used to be use standards anything was ok. Now: Need to setup compatible, to present a single point of view to our users. To first order a site can implement whatever it wants, so long as it passes the test, but being pragmatic if you want our help , we have to focus on a few core technologies. We are going to see cuts on both projects in the coming years. That’s why we are looking at glite. May be better off that has a much broader community update. Already setting up tools to be able to integrate APEL with RUS (at Manchester). Can’t have expertise in everything. Q: Compute stack seems to be diversifying AR: We can track the other work and leverage these technologies. One HPCC sig site looking at ARC. Duncan: The title of the doc is NGS to NGI roadmap. Talks about consolidating to one solution, talks about shibboleth, also talks about UK solutions. 1 To support uk users 2 To interoperate with an influence EGI 3 To integrate with GRIDPP where its advantageous.
NGI is not just about NGS and GridPP turning into a single thing. NGI operations group chaired by John Gordon. EGI is not everything to the UK , but it doesn’t make sense to have two grids, so need to work closer together. Should continue working with US, Terra Grid.
DW: Need to make certain we continue to support external groups. Not going to consolidate around single tools unless there is a very good reason.
DW: The UK will continue to have its own voice in this. Cannot let our voice be subsumed by a small community, we represent the whole research community. Document needs to include the OGF work.
Still a draft document.
AR: Tries to define a vision irrespective of funding. Then even if the NGS disappears next April , there is still a plan. Mitigating risk completely is not something you can do. What happens if no gridpp4, no NGS4, .... DW: Need to define the services so where funding is constrained, can decide what they want to do. When they write the NGS4 proposal in the next 6 months, will have to define what they get for different levels of funding. SY: What are the timescales for this doc. DW: Completed version by next Management board. (Sept/Nov?) AR: GridPP are already making a written response to the doc.
