OSG Newsletter, January 2010
ATLAS Analysing Real Data After LHC Start-up
ATLAS (A Toroidal Lhc ApparatuS?) is one of the four particle physics experiments located at the Large Hadron Collider at CERN. With the first sustained operation of the LHC in December, ATLAS has finally taken substantial amounts of real collision data—ultimately at record-setting energy (2.36 TeV?).
Subsequently, in late December and early January, U.S. ATLAS made heavy use of its sites on the OSG to perform reconstruction and initial analysis of this data. During these processing runs, the ATLAS Tier 1 Center at Brookhaven National Lab, along with 5 Tier 2 sites (and a growing number of Tier 3s) typically ran 10-13,000 jobs at a time, with a maximum throughput of almost 50,000 jobs over a 24-hour period.
Thanks to the OSG, this work was accomplished quickly and with very high efficiencies (~99%). ATLAS plans to perform another reprocessing run (which means re-analysing the December data with refined calibration information) before the LHC restarts in February.
~ John Hover
Production Report from CMS
CMS continues to shift focus from exercising the computing system at full load to analyzing the roughly 15 terabytes of real collision data from the 2009 run. The emphasis is on checking how well Monte Carlo simulations match the data collected by the detector, and so far they are in fantastic agreement! At the well-understood center-of-mass energy of 900 GeV, the detector energy loss, particle spectra, and reconstructed masses and widths for well-known particles, such as lambdas and K^0 shorts (to name but a few quantities of interest), have matched up well to expectations.
~ Burt Holzman
Native Packaging VDT Work
The VDT team at the University of Wisconsin-Madison has been working hard to be able to ship the OSG Software Stack as native packages (RPMs and Debian packages) in addition to the existing Pacman distribution. There has been strong demand from VOs and users for native packages because they would interact well with the OS distribution mechanisms that site administrators already use.
The work has been challenging because we are becoming experts in two native packaging systems, and because much of our internal infrastructure for producing the OSG Software Stack relies on features in Pacman that do not match well with native packaging systems.
Right now, the VDT team is focusing on a subset of the VDT needed by the LIGO experiment, which has the most pressing needs for native packaging. After this, we hope to provide native packages, first for glexec (because it must be installed on a local disk), and then software needed on the worker nodes. Once that is done, we’ll gradually increase the set of available native packages.
We anticipate continuing to support Pacman installations for at least a while. Users have already installed software with Pacman, and should not need to re-install from scratch just to get a new style of packaging. We’ll support Pacman for the foreseeable future.
We look forward to working with all of you as we proceed with support for native packages.
~ Alain Roy
Early LHC Measurements from the Large Hadron Collider Depend on the Distriburted Computing Contributions of the Open Science Grid
In November and December 2009, the LHC delivered over one million collisions to the four detectors at energies of 900 GeV and 2,360 GeV, please view the US LHC Blog Entry. Many physicists having been poring over the data during and since that data run. Early measurements are already being validated, reviewed, and presented by the experiments. These measurements look at detection and distribution of known particles, masses, and distributions; they check the performance of each of the millions of channels of the detector and compare the actual data to that from the years of simulation.
You can catch some of the excitement locally at the US LHC Blog entry from Edgar Carrera of Boston University. We give two publicly available plots to whet your apetitite!
First physics distributions: The green band represents the measurement uncertainty (Courtesy of Fermilab Today, Don Lincoln)
~ OSG Executive Team
iSGTW Survey: January 2010
As you know, the SGTW (Science Grid This Week) was created by us in OSG, then grew to include the “i” in partnership with EGEE and GridPP? in Europe. iSGTW now has 5,684 subscribers. We need your feedback to continue to make this publication worthwhile and useful.
Please take a few moments to fill out the survey. The editors welcome opinion pieces as well as news. Let’s make sure we keep articles coming and the many subscribers informed and up-to-date.
From the Executive Director
Your many contributions were key to the comprehensive annual report that I submitted to the DOE this week. Well done! Do take a moment to read some of the OSG 2009 Annual Report.
This month’s focus in several areas is usability, scalability, usability, scalability…. Usability was a theme echoed throughout the ID Management Workshop, please view the report. If you want to know more and want to help, come to the All Hands Meeting in March!
Another new OSG satellite project—the Advance Network Initiative (INI) research project at DOE funded by ARRA—has kicked off with work at UCSD. The goals and initial activities will make use of a local testbed in collaboration with SDSC and then with NERSC, while the wide-area 100 Gigabit fabric is being built.
And don’t forget to register for the All Hands Meeting. See you there!
December ATLAS & CMS Thumbnails
Recent change in profile of US LHC usage on OSG reflects increased attention to analysis jobs and temporary reduction of large scale simulation jobs.
Documents Modified in the Last 20 Days
OSG-doc-# Title Author(s) Topic(s) Last Updated 931-v2 OSG-ESnet Joint Identity Management Workshop report Mine Altunay et al. Security Plans
15 Jan 2010 932-v2 OSG Annual Report to DOE (December 2009) Paul Avery et al. Reports
15 Jan 2010 930-v2 Vignettes to ASCR Miron Livny et al. Outreach
11 Jan 2010 813-v2 The Value of OSG Executive Board Reports
08 Jan 2010 887-v1 Year 3 Metrics Brian Bockelman Metrics
08 Jan 2010 928-v0 RSV (Draft) Rob Quick Agreements
06 Jan 2010 927-v0 CA Distribution Rob Quick Agreements
06 Jan 2010 926-v0 MyOSG Service Level Agremeent Rob Quick Agreements
06 Jan 2010 925-v0 BDII Service Level Agreement Rob Quick Agreements
06 Jan 2010