Скачать презентацию U S ATLAS Computing Facilities Overview Bruce G Скачать презентацию U S ATLAS Computing Facilities Overview Bruce G

2032c45efe6e46ccc94de6638e47fb3e.ppt

  • Количество слайдов: 19

U. S. ATLAS Computing Facilities (Overview) Bruce G. Gibbard Brookhaven National Laboratory US ATLAS U. S. ATLAS Computing Facilities (Overview) Bruce G. Gibbard Brookhaven National Laboratory US ATLAS Computing Advisory Panel Meeting Argonne National Laboratory October 30 -31, 2001

US ATLAS Computing Facilities Mission T Facilities procured, installed and operated q …to meet US ATLAS Computing Facilities Mission T Facilities procured, installed and operated q …to meet U. S. “MOU” obligations to ATLAS z z z q Direct IT support (Monte Carlo generation, for example) Support for detector construction, testing, and calibration Support for software development and testing …to enable effective participation by US physicists in the ATLAS physics program! z z Direct access to and analysis of physics data sets Simulation, re-reconstruction, and reorganization of data as required to complete such analyses B. Gibbard US ATLAS Computing Facilities 30 October, 2001 2

US ATLAS Computing Facilities Overview T A Hierarchy of Grid Connected Distributed Resources Including: US ATLAS Computing Facilities Overview T A Hierarchy of Grid Connected Distributed Resources Including: q Tier 1 Facility Located at Brookhaven – Rich Baker / Bruce Gibbard z q Operational at ~ 0. 5% level 5 Permanent Tier 2 Facilities (to be Selected in April ’ 03) z 2 Prototype Tier 2’s selected earlier this year and now active s s q Tier 3 / Institutional Facilities z z q Indiana University – Rob Gardner Boston University – Jim Shank Several currently active; most candidate to become Tier 2’s Univ. of California at Berkeley, Univ. of Michigan, Univ. of Oklahoma, Univ. of Texas at Arlington, Argonne Nat. Lab. Distribute IT Infrastructure – Rob Gardner z z z US ATLAS Grid Testbed – Ed May HEP Networking – Shawn Mc. Kee Coupled to Grid Projects with designated liaisons s s PPDG – Torre Wenaus Gri. Phy. N – Rob Gardner i. VDGL – Rob Gardner EU Data Grid – Craig Tull B. Gibbard US ATLAS Computing Facilities 30 October, 2001 3

Evolution of US ATLAS Facilities Plan T In Respond to Changes or Potential Changes Evolution of US ATLAS Facilities Plan T In Respond to Changes or Potential Changes in q Schedule q Requirements/Computing Model q Technology q Budgetary Guidance T Changes in Schedule q LHC start-up projected to be a year later, 2005/2006 2006/2007 q ATLAS Data Challenges (DC’s) have, so far, stayed fixed z z z DC 0 – Nov/Dec 2001 – 105 events – Continuity Test DC 1 – Feb/Jul 2002 – 107 events ~ 1% DC 2 – Jan/Sep 2003 – 108 events ~ 10% - a serious Functionality/Capacity exercise B. Gibbard US ATLAS Computing Facilities 30 October, 2001 4

Changes in Computing Model and Requirements T Requirements Defined by International ATLAS Computing Model Changes in Computing Model and Requirements T Requirements Defined by International ATLAS Computing Model T Nominal model and requirements for a Tier 1 (Expect there to be ~6) q Raw ESD/AOD/TAG pass done at CERN, result shipped to Tier 1’s q TAG/AOD/~25% of ESD on Disk, Tertiary storage for remainder of ESD q Selection passes through ESD monthly q Analysis of TAG/AOD/Selected ESD/etc. (n-tuples) on disk within 4 hours by ~200 users requires … B. Gibbard US ATLAS Computing Facilities 30 October, 2001 5

Changes in Computing Model and Requirements (2) T Revised model and requirements for a Changes in Computing Model and Requirements (2) T Revised model and requirements for a Tier 1 (under consideration) q Raw ESD/AOD/TAG pass done at CERN, result shipped to Tier 1’s q TAG/AOD/33% of ESD on Disk at each Tier 1 (3 sites in aggregate contain 100% of ESD on Disk) q Selection passes through ESD daily using data resident on disk locally and at 2 complementary Tier 1’s q Analysis of TAG/AOD/All ESD/etc. (n-tuples) on disk within 4 hours by ~200 users requires … B. Gibbard US ATLAS Computing Facilities 30 October, 2001 6

Comparing Models T All ESD, AOD, and TAG data on disk greatly speeds/improves analyses Comparing Models T All ESD, AOD, and TAG data on disk greatly speeds/improves analyses q Enables one day selection passes (rather than one month) and reduces the tape requirement imposed by selection processing – better/faster selection q Allows navigation of individual events (for all processed, but not Raw, data) without recourse to tape and associated delay – more detailed/faster analysis q Avoids contention between analyses over ESD disk space and the need for complex algorithms to optimize use of that space – less effort for better result T But there are potentially significant cost and operational drawbacks q Additional disk is required to hold 1/3 of ESD q Additional CPU is required to support more frequent selection passes q It introduces major dependencies between Tier 1’s q It increases sensitivity to performance of the network and associated Grid middleware (particularly when separate by a “thin” pipe across an ocean) T What is optimal for US ATLAS computing? B. Gibbard US ATLAS Computing Facilities 30 October, 2001 7

Changes in Technology T No dramatic new technologies T Previously assumed technologies are tracking Changes in Technology T No dramatic new technologies T Previously assumed technologies are tracking Moore’s Law well T Recent price performance points from RHIC Computing Facility q CPU: IBM procurement - $33/SPECint 95 z z z q 310 Dual 1 GHz Pentium III nodes @ 97. 2 SPECint 95/Node Delivered Aug 2001, now fully operational $1 M fully racked including cluster management hardware & software Disk: OSSI/LSI procurement - $27 k/TByte z z z 33 Usable TB of high availability Fibre Channel RAID 5 @ 1400 MBytes/sec Delivered Sept 2001, first production use this week $887 k including SAN switch T Strategy is to project, somewhat conservatively, from these points for facilities design and costing q Using somewhat longer than the observed <18 month price/performance halving time – detailed capacity & costing will be presented by Rich Baker B. Gibbard US ATLAS Computing Facilities 30 October, 2001 8

Changes in Budgetary Assumptions T Assumed Funding Profiles ($K) T For revise LHC startup Changes in Budgetary Assumptions T Assumed Funding Profiles ($K) T For revise LHC startup schedule, new profile is better q In new profile, funding for each year generally matches or exceeds that for one year earlier in the old profile. q Funds are more effective when spend 1 year later (Moore’s Law) T For ATLAS DC 2 which stayed fixed in ’ 03, new profile is worse q Hardware capacity goals of DC 2 cannot be met q Personnel intensive facility development may be up to 1 year behind q Again, Rich Baker will discuss details T Hope/expectation is that another DC will be added allowing validation of more nearly fully developed Tier 1 and US ATLAS facilities Grid B. Gibbard US ATLAS Computing Facilities 30 October, 2001 9

Capacities of US ATLAS Facilities for Nominal Model B. Gibbard US ATLAS Computing Facilities Capacities of US ATLAS Facilities for Nominal Model B. Gibbard US ATLAS Computing Facilities 30 October, 2001 10

Revised US ATLAS Tier 1 Model T Stretch out of LHC startup schedule combined Revised US ATLAS Tier 1 Model T Stretch out of LHC startup schedule combined with DOE late funding ramp-up allows for significantly improved US ATLAS Tier 1 facility in ‘ 07 (rather than ‘ 06) while staying within budget (Unfortunately it does not help for DC 2) T It is based on the Revised International ATLAS Model with augmentation to address operational drawbacks q Increase disk to hold 100% of ESD z z Removing dependency on other Tier 1’s Reducing dependency on network across the Atlantic q Add sufficient CPU to exploit highly improved data access q Retain tape storage volume of one STK silo, reduce tape I/O bandwidth to only that required in new model (Selection from disk not tape) B. Gibbard US ATLAS Computing Facilities 30 October, 2001 11

Revised US ATLAS Tier 1 Model (2) T Impact on Overall US ATLAS Computing Revised US ATLAS Tier 1 Model (2) T Impact on Overall US ATLAS Computing Model q The high availability of the complete ESD set at the Tier 1 makes possible more prompt and detailed analyses by users at coupled Tier 2 and Tier 3 sites as well as those directly running at the Tier 1 q Increased CPU capacity to exploit this possibility at these site is desirable and may be feasible given the 1 year delay in delivery date, but such an expansion remains to be studied q Exploitation of this capability would increase the network load between Tier 2/3 sites and the Tier 1 and thus the network requirement but the again the added year should help and further study is required T Conclusions q It is currently our intent to make this revised plan the default US ATLAS Tier 1 plan and to determine what changes in the overall US ATLAS facilities plan should and can efficiently follow from this B. Gibbard US ATLAS Computing Facilities 30 October, 2001 12

Capacities of US ATLAS Facilities for Revised Model B. Gibbard US ATLAS Computing Facilities Capacities of US ATLAS Facilities for Revised Model B. Gibbard US ATLAS Computing Facilities 30 October, 2001 13

Tier 1 Ramp-up Profile * DC 2 B. Gibbard US ATLAS Computing Facilities 30 Tier 1 Ramp-up Profile * DC 2 B. Gibbard US ATLAS Computing Facilities 30 October, 2001 14

B. Gibbard US ATLAS Computing Facilities 30 October, 2001 15 B. Gibbard US ATLAS Computing Facilities 30 October, 2001 15

B. Gibbard US ATLAS Computing Facilities 30 October, 2001 16 B. Gibbard US ATLAS Computing Facilities 30 October, 2001 16

STATUS of Tier 1 Facility Evolution T Goal of Planned Technical Evolution in FY STATUS of Tier 1 Facility Evolution T Goal of Planned Technical Evolution in FY ’ 01 was to Establish US ATLAS Scalability & Independence (from RCF) q Users Services – 100 registered users z z q Infrastructure Services z z q NIS, DNS, etc. servers SSH gateways SUN/Solaris Services z z q Accounts, passwords, CTS, etc. Documentations Server NFS disk AFS Service z z AFS servers AFS disk q Network q HPSS Service z z Server Tape/Cache disk B. Gibbard US ATLAS Computing Facilities 30 October, 2001 17

B. Gibbard US ATLAS Computing Facilities 30 October, 2001 18 B. Gibbard US ATLAS Computing Facilities 30 October, 2001 18

STATUS of Tier 1 Facility Evolution T Goal of Planned Technical Evolution in FY STATUS of Tier 1 Facility Evolution T Goal of Planned Technical Evolution in FY ’ 01 was to Establish US ATLAS Scalability & Independence (from RCF) q Users Services – 100 registered users z z q Accounts, passwords, CTS, etc. Documentations Infrastructure Services z NIS, DNS, etc. servers z q SSH gateways SUN/Solaris Services z Server * z q NFS disk * AFS Service z AFS servers * z AFS disk * q Network * q HPSS Service z Server * z Tape/Cache disk * B. Gibbard US ATLAS Computing Facilities 30 October, 2001 19