Attendees
Ray Plante
Jacek Becla
Cristina Beldica
Chris Smith
Don Dossa
Andy Connolly
Tim Axelrod
Jeff Kantor

Phone Lines for Videocons for WG meetings ­ Jeff
Iain needs IP numbers of participants for firewall reasons.  Whoever patches
in the phone needs to learn how from Iain.  If no one available to patch in
just revert to telephone.

Debrief of SLAC Director's Review - Tim/Jeff
Review was ³friendly² review panel from SLAC, LBNL, UA, Lick.  Principal
findings were risk in sensor development, overly tight schedule, more detail
needed in cost estimates, science program does not clearly demonstrate
advantage/need for high etendue.  We assume that these items will need to be
addressed before Director will give a ³go².  Review panel will issue written
report shortly.

MREFC Proposal ­ Jeff
Don Sweeney met with Mark Coles, responsible for NSF MREFC programs.  NSF
has new process for MREFC that more closely mirrors DOE process.  Coles
indicated that certain parts of process may take longer, which may imply we
will not receive construction funding by 2009.

DC1 Planning update - Tim/Ray/Cristina
New DC1 telecon on Tuesdays at 11:00am PST, starting next week.  For DC1 we
need to decide how to reduce throughput (smaller images or fewer images?).
Could use different scaling factors in different parts of DC1, not all need
to be 2%.  Can proceed empirically, i.e. turn each ³knob² up until it hurts,
then back off. Some limits, i.e. Ingest, will be known by prior tests before
DC1.  Will do DC1 at CCD-parallel level.

Middleware API status - Ghaleb/Ray/Marcus
Ghaleb will reduce LSST direct involvement, due to LDRD/other program
responsibilities, continue working in provenance area as available.  Ray
will assume chair of WG.  Charge 2 weeks ago was to evaluate/clean current
Middleware API, ready to go to the next level.  Jacek worked Ingest, Ray has
changed Pipeline Execution, working on Policies.  This week, assemble list
of specific robustness, sequence diagrams out of skunk works.  Tim and Jeff
talk to Doug this week to set expectations for meeting.

Camera DM Interface Update - Tim/Don
Discussions with SLAC people, Don, Tim, Jeff.  Mike Huffer has put proposal
out: system extends from mountain to the base, the client interface will be
at the base.  One issue is lots of software/hardware complexity at camera
end.  Uses Infiniband, offers services.  Crosstalk location is still up in
the air, proposal does it in the interface.  This is appealing from
performance standpoint. Issue is that doing this in the camera makes them
part of the provenance requirement, and may be inflexible if there is xtalk
between rafts.  Raw data will be sent anyway, will need to be re-ordered
after xtalk correction for rest of image processing.  Don still looking at
solutions for IB over distance.

File systems and replication (GFS status, xrootd) - Jacek/Ray
Chris Cribbs has installed, is now starting to work with it.  GFS has been
built by UCD postdoc, should be available as open source by summer.  Tony
Tyson, Jeff working on legal agreement allowing us to evaluate it at NCSA,
UCD, LCO.



_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data

Reply via email to