50 likes | 164 Views
Analysis Model sub-committee status report. Status. We are rapidly converging on an analysis model See next slides for an overview We are working on a draft requirements document We will provide this to the full committee next Tuesday
E N D
Status • We are rapidly converging on an analysis model • See next slides for an overview • We are working on a draft requirements document • We will provide this to the full committee next Tuesday • The requirements are heavily annotated to describe possible implementations and development implications • Supporting documentation and ‘Analysis Scenarios’ will be provided with the requirements doc • Issues outside the strict definition of an analysis model are left open • No requirements on file formats (Objy/Kanga) • Overlaps with Resource and AWG management issues will need to be worked out during the workshop David Nathan Brown 2
Basic Outline of the Requirements • BaBar will support 2 data content formats: The Mini and an upgraded (New) Micro • The New Micro will be readable both by a Framework input module and from the Root/CINT prompt • Eric Charles has a working demonstration of dual-read micros • Framework read must be very fast (~1KHz/event) • Requires improving Beta • Both formats will allow customization by users/AWGs • Add Composites and user-specific data • Allow choice of which lists and what level of detail are stored • Both formats will be fully accessible from all Tier-A sites • Small Mini samples will be readily accessible • Large Mini samples (many files) may take up to 1 month to access • A ‘minimal’ copy of the New Micro will be readily accessible • ~1KByte/event • AWG-specific skims will be produced regularly • Output may be in either format, depending on physics needs • Output may be pointer or deep copy, depending on size David Nathan Brown 3
Implementation • The requirements must be implementable within reasonable development resource limits • Proof-of-principle or better exists in most cases • General agreement about the rest • Implied by the requirements: • The implementation should be easy to use and flexible so that people will naturally find the formats useful • The formats provide an overlapping range of options for accessing data and creating customized work samples • AWGs and individual users can figure out what works best for them within well defined resource limits • Improve the infrastructure so that resources are freed up • We are still working out specific analysis strategies • Explicit scenarios will be presented with the requirements David Nathan Brown 4
Migration • We need to implement the requirements in ~1 year • The migration should not unduly disrupt existing analyses • Some disruption is inevitable as the cost of change • Resources must be carefully allocated • Both old and new formats will need disk space • Developers must be given clear priorities • We need both carrots and sticks to encourage users to migrate • Carrots should be the better access, better data content, more efficient use of resources (cpu + disk space) in the new model • Sticks should be firm dates to recycle space used by old formats • We need a detailed and believable migration plan • Some discussion of this should happen in the workshop David Nathan Brown 5