130 likes | 334 Views
EGEE. EGEE - The Network Sector. EGEE. Operation of EGEE relies upon Geant and NRENs A priori EGEE will make demanding requirements: Topology of Traffic loads Value added services over and above best efforts IP …etc…. EGEE - The Network Sector WHY ?.
E N D
EGEE EGEE - The Network Sector
EGEE • Operation of EGEE relies upon Geant and NRENs • A priori EGEE will make demanding requirements: • Topology of Traffic loads • Value added services over and above best efforts IP • …etc….. EGEE - The Network Sector WHY ? • Inconceivable that HEP should not influence EGEE • Inconceivable that UK would not wish to be a partner • EU mandates close co-operation with Geant • Is one of the things everyone is happy for the UK to coordinate (at present)
Geant’ EGEE EGEE First task is to define what fits legitimately in scope of EGEE
…………….. NREN UKERNA Geant Policy Committee DANTE DFN NL TERENA UK …………….. IT EGEE
EGEE • Additional capacity requirements: • It is possible that upgrades to INTERNATIONAL links are required specifically for the Grid infrastructure. Case will have to be made from EGEE. • Requirements capture and modelling • Performance monitoring & Diagnostics • Need integrated End-to-end performance monitoring and diagnostic services ( Instrumentation throughout network (core and edges), Diagnostic engines) • GOPS (Grid Ops centre) will need access to comprehensive tools in addition to the above
EGEE • “Production” services for control and allocation of network resources (i.e. Bandwidth, Qos, Secure Virtual Networks, Extended VLANs….) • User interface to scheduling and reservation services • e2e control plane • provisioning mechanisms • Interface to Grid middleware • Scalability • ..of course AAA • Data Transport: • Key to Grid operations. • Must ensure results of current projects to achieve > Gbit/s transport become usable by Grid centres.
EGEE Next steps • Convene working group which represents stakeholders • Define scope (Ensure boundaries and relation between Geant infrastructure and EGEE infrastructure funding are agreed and clear) • Start to prepare a work package for proposal…… • Hold meetings
GridPP-2 Middleware DEVELOPMENT & SUPPORT
Suggested policies • Don’t try to cover everything (a la EDG) • Support a limited number of “middleware” areas properly • Support those which are “Mission critical” to HEP (e.g. Data Management) • Support areas where we have a clear leadership role
Suggested policies • GridPP-2 should deal with “logical centres” • Not sensible to deal with many individual institutes at the level of 0.5 posts ..etc.. (this puts the onus on coordinating a sector on GridPP-2 resource management which is silly) • Instead a “logical centre” agrees to deliver on a specific area – against several posts. • Logical centre = { critical mass, credibility, expertise…} • Can be grouping of institutes. But this is invisble to GridPP-2 at the level of resource management. • Perhaps should have inter disciplinary / industry agle
Much of current “middleware R&D in work packages” will evolve to “operations and support” • Taper down of effort in middleware • GridPP should move from development to exploitation • Should take progressively more from external “suppliers”
Status of deliberations • SLOW • Document just drafted with first thoughts => PMB • Next it goes to • TB : axiomatic that people involved in work say what they think are logical next steps in middleware form HEP point of view • EB: What experiments want of GridPP-2 must shape this sector • Middleware Workshop to be organised • Day 1: present in detail what we have all done • Evening 1: Stitch up in pub • Day 2: Discuss where we should go (to justify prior step)