r2 - 08 Mar 2011 - 13:07:44 - TorreWenausYou are here: TWiki >  AtlasSoftware Web > AgencyReview201103PSC

Physics Support and Computing Materials for the March 2011 DOE/NSF Review of the U.S. ATLAS Operations Program


Physics Support and Computing materials are gathered here for the March 2011 DOE/NSF review of the U.S. ATLAS Operations Program. The review website is here.

U.S. ATLAS Operations Program Manager's Review, January 27-28, 2011

In April a U.S. ATLAS Operations Program Manager's Review was held at BNL, chaired by Steve Vigdor. Agenda is here, password usatlasmr. This review included detailed talks on Computing Facilities and Software which are not covered by dedicated talks in the agency review. Reviewing their content is recommended to gain more detailed information in these areas than is available in the agency review talks. Some highlights from these presentations are included in the supplementary slides of Torre's overview talk in the agency review.

Recommendations to PS&C

Recommendations from the operations program manager's review have not yet been received.

DOE/NSF Review of the U.S. ATLAS Operations Program, May 2010

PS&C materials for the May 2010 operations program review are on this page.

Recommendations to PS&C

Here follow the reviewers' recommendations in the PS&C area and a response (in green).

Evaluate the impact of the jamborees and forums and make adjustments as necessary. Engage U.S. ATLAS collaborators to determine if a more cohesive collaboration of U.S. participants is desired.

Up until the summer of 2010, the Jamborees and U.S. Weekly meetings (focusing on the Forums) were geared towards the startup phase of the experiment. The consensus is that these have been quite useful and successful. However, attendance at both the Jamborees and the U.S. weekly meetings has been declining recently. With the arrival of data, the needs and desires of the U.S. ATLAS physicists have changed significantly. Now that many ATLAS physics and performance results are being produced, U.S. analysts must work even more closely with the ATLAS physics/performance groups than they had previously done. Also there is less need for technical analysis tutorials; most analysts are quite experienced and the documentations (especially the workbook) are well developed. In response, we are considering various changes to the analysis support organization. It is expected that these changes will lead to a smaller number of Jamborees and a reduction in frequency of the "Weekly" meeting. These Jamborees and regular meetings are expected to be much more targeted on topics specific to U.S. analyzers. It is also expected that there will be a move to a looser organization based more on analysis groups being fostered at an ASC, or elsewhere. The U.S. ATLAS Fellows program will undoubtedly play a key role here.

With respect to the question of whether U.S. ATLAS collaborators prefer a more coherent U.S. Collaboration: As part of the search for the new Analysis Support Manager, U.S. ATLAS conducted a poll on the analysis support needs from the member institutes. The poll found that such cohesion would be particularly helpful for the smaller university groups. Thus, Analysis Support will put particular emphasis on engaging such groups.

Pursue the planned cost-benefit analysis of computing resources taking in the needs of the ATLAS U.S. physicists and U.S.-based research as well as meeting international commitments and report the results at the next U.S. ATLAS operations review.

During 2010 PS&C management together with our Tier 2 PIs completed a comprehensive cost-benefit analysis of computing resources from the present through 2016. The analysis was made in the context of best-estimate resource requirements for ATLAS computing over this period, with the U.S. share estimated at the current value of 23%. U.S. resource planning has always included a 20% component dedicated for the use of U.S. physicists, when budgets permit, over and above the resources pledged to ATLAS, and this component was included in the analysis. The analysis involved each Tier 2 making a close study of its present and future costs, and negotiating with consortium institutions for further local contributions to supplement future project funding. The Tier 2s formalized their plans, costs, local resource/funding contributions, deliverables, and broader contributions to U.S. ATLAS and ATLAS (such as expertise, technical development and R&D contributions) for the next 5 years in proposals for their continued participation as U.S. ATLAS Tier 2 centers out to 2016. These proposals on the foundation of the cost-benefit analysis and resource requirement estimations formed the core of the PS&C component of the NSF Cooperative Agreement proposal for 2012-2016, the target funding level of which is sufficient to fund our NSF Tier 2s at the levels estimated as required from the analysis. The cost-benefit analysis results and the facility plans documented in these proposals will also be used in computing facility funding and resource placement decisions going forward. The Tier 2 proposals and an explanation of the resource requirement estimation can be found on this page.

U.S. Distributed Facility cost/benefit analysis

During fall 2010 PS&C conducted a comprehensive cost/benefit analysis across the facilities, with particular focus on the Tier 2s. The analysis was done to

  • guide resource allocation decisions to maximize value for dollar
  • provide a current, quantitative basis for developing budgets required to deliver capacity requirements
  • plan the evolution of the Tier 2 complex over the coming years
  • revisit the interest of participating institutes going foward, and what local contributions they are able to bring
  • examine whether new institutes should be added to the complex, and if so, how
  • provide a detailed basis for requesting renewal of the NSF-supported Tier 2 program in the NSF Cooperative Agreement Proposal submitted December 2010

Information on the analysis is on this page:

  • proposals developed by Tier 2s for their continued participation
  • Proposal from University of Illinois/NCSA for participation as a new Tier 2 site. Their proposal was accepted for inclusion in the NSF Cooperative Agreement Proposal, with UI/NCSA planned to be included as a new MWT2 consortium member.
  • Detailed explanation of how computing requirements estimates for 2011-2016 were made.

The cost/benefit analysis has been used extensively for capacity and budget planning for FY11-FY16.

Prioritization of PS&C activities

During early 2011 PS&C conducted a comprehensive review and prioritization of activities across the project. Each area manager developed an itemized, prioritized list of activities, with descriptions of the importance of the activity and the impact the loss of the effort would have. Prioritization was done in three priority tiers. Prioritizations were reviewed by project management and iterated on. Activities at level 3 prioritization (the lowest) were surveyed for issues to address and appropriateness of continuing the activity. Activities including level 3 were found to be solid, well motivated, and delivering strong value to U.S. ATLAS and ATLAS, reflecting in part the benefit of a similar analysis last spring which was followed by corrective actions in a number of areas.

Present understanding of budget expectations and their consequences for the project do not indicate that cuts in project activities and personnel will be required in the near term. However if that situation changes the prioritization will guide decisions on making necessary cuts. Present budget restrictions reduce the software effort in 2013 and out years, should these reductions take effect, again the prioritization will guide the decisions.

Addressing budget scenarios

ATLAS has been very actively reviewing and modifying its computing model for 2011-2012 running, in light of resource restrictions due to tight budgets in funding agencies, an expected substantial rise in data volumes and processing loads coming with increases in integrated luminosity and pileup, and a strong physics case for running at a high trigger rate (300-400Hz). In this process ATLAS has been converging on a new profile for required computing capacity 2011-2013. In the U.S. we developed in last fall's facility cost/benefit analysis our own estimate of the required capacity profile based on existing ATLAS estimates (which did not go beyond 2011 at the time) and the prevalent ATLAS computing model. These estimates formed the basis of our initial budget planning for FY11+. Budget guidance representing a substantial reduction in the PS&C budget was received at the same time ATLAS produced a new capacity requirements profile that is substantially less than we anticipated based on the ATLAS computing model -- the reductions reflected an aggressive change of model on the part of ATLAS to address funding realities and physics requirements.

We have addressed PS&C budget scenario targets by updating our facility budgets to the new ATLAS capacity profile (at our 23% share), and achieving the additional cuts required to come within targets by making reductions elsewhere in the project (software), with specific cuts to be determined based on the prioritization we have done and the circumstances prevalent when the cuts have to be applied.

Useful information on the present state of the US ATLAS facility and the expectations and plans for 2011 and beyond can be found in the talks at the US ATLAS facilities meeting co-located with the OSG All Hands Meeting earlier this week. See the agenda.

Tier 3s

The majority of Tier 3 sites in the U.S. are 'Tier3g' sites which are relatively autonomous from grid infrastructure, but are nonetheless integrated into ATLAS Distributed Computing services. The U.S. is a leader in ATLAS Tier 3 planning, implementation and support and leads the technical coordination of ATLAS Tier 3s (Doug Benjamin, Duke). The U.S. has led the development of detailed documentation on Tier 3g setup and usage (found here, but on the ATLAS protected wiki).

Current information on U.S. ATLAS Tier 3 activity can be found on the agenda for this week's US ATLAS Facilities Meeting co-located with the OSG All Hands Meeting. See the Tuesday morning session, and the link to the common ATLAS/CMS Tier 3 session on Tuesday afternoon. Of particular interest: D. Benjamin's U.S. ATLAS Tier 3 overview, and S. Willocq's talk on ATLAS Tier 3 analysis experience.

Major updates:
-- TWikiAdminGroup - 22 Mar 2018

About This Site

Please note that this site is a content mirror of the BNL US ATLAS TWiki. To edit the content of this page, click the Edit this page button at the top of the page and log in with your US ATLAS computing account name and password.


Powered by TWiki
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback