Atlas CAP Closeout

1
Atlas CAP Closeout

• Thanks to all the presenters for their work in
  preparing documentation, presentations, web
  sites, and demos!
• Personal thanks for tolerating my inability to
  travel

2
Project Overall

• The excitement is building real data is now
  only a year away!
• U.S. Atlas (and International Atlas) have made
  good and encouraging progress in the past 12
  months
• Budgets remain very tight (as always)

3
Management (1)

• Overall, the management of the Software
  Computing part of the research program appears
  healthy...
• Leadership is clearly engaged within the U.S.
  project, within International ATLAS, and in
  related grid projects

4
Management (2)

• (Formal) Change Control has made progress in the
  last 12 months
• Process is being used
• Triggers are better defined
• ...but there is still room for improvement so
  that it becomes a useful management tool.
• Once again multiple changes were rolled into an
  aggregate omnibus change, which is not a
  recommended approach.
• Recommendation Utilize Change Control more
  formally, even allowing documented requests to
  appear earlier that may or may not be approved,
  or may be deferred.

5
Management (3)

• For FY2007, an adequate management reserve is
  being maintained. For FY2008, it is ephemeral,
  in that the computing project needs are of the
  scale of the full reserve (i.e. there is no
  reserve).
• The leadership did state that the currently
  unfunded (held in management reserve) components
  would have to compete for the funds (hence cannot
  be assumed).
• Full Dress Rehearsal is a concern we encourage
  management to review carefully goals readiness,
  and interact with International ATLAS if it
  appears too ambitious (partitioning may be
  needed).

6
Facilities (1)

• There has been healthy growth in all aspects of
  facilities this year
• Tier 1 capacity growth
• Tier 1 staff growth
• Tier 1 network bandwidth upgrades
• Selection of final 2 tier 2s
• SC 4 was a success. Achieved 200 MB/s sustained
  (average)
• Impressive U.S. ATLAS contributed 28 of
  distributed event production the BNL Tier 1 is
  recognized as doing very well.
• There are no procurement issues in 2007
• Starting in 2008-9, planned facilities growth is
  dependent upon consumption of 50 of the
  management reserve.
• In 2010, even that reserve will not be sufficient
  to maintain the goal of U.S. ATLAS additional
  capacity (beyond International ATLAS commitments)

7
Facilities (2)

• There is no ATLAS specific cybersecurity plan.
  The Tier 1 plans appear reasonable, but U.S.
  Atlas is dependent upon the Tier 2 sites (and
  OSG) to address the distributed cybersecurity
  problem.
• We endorse the goal of identifying cybersecurity
  manpower for ATLAS, and believe in the current
  culture it will become essential (even mandatory)
• We encourage the project to finalize plans with
  BNL for deployment space for 2009 and beyond.

8
Distributed Computing (Grid)

• Impressive progress on Panda in the last 12
  months, moving from a relatively new prototype (2
  months old) to a rapidly maturing production
  tool.
• The next 12 months will be key as Panda makes
  opportunistic use of OSG resources.
• Engagement level w/ OSG has significantly ramped
  up.
• Full Dress Rehearsal is a concern

9
Software (1)

• U.S. Atlas is involved at a healty level in event
  generators, infrastructure for tracking,
  calorimeter, muons, and monitoring
• We endorse the plans for a graphics programmer,
  and expect the need to continue at some level
  beyond 2 years
• Generator support may be needed in the out years
  (currently drops to 0 in 2008)
• Development of a Derived Physics Data format
  (standard) is encouraged, lest multiple formats
  emerge

10
Software (2)

• The demos were very good!
• pathena demo convincing, showing that the system
  is now a reality
• status browsing harder for non expert to follow
  more complex, lots of detail

11
Analysis Support

• Great progress made in 2006
• better defined than a year ago
• many plans from a year ago have been realized
• Distributed model looks promising, good way to
  get more people involved
• Jamborees are popular, successful
  congratulations on enthusiasm generated
• Services of ASG are not being taken advantage of
  fully yet. If new staff is needed for the
  Jamborees, then proceed quickly. Otherwise, pace
  to growth in demands for support.
• Need to address support for people at CERN