September 13, 2012
The discovery of the Higgs boson is a major scientific achievement, the culmination of 48 years of dedicated effort by the global High Energy Physics (HEP) community.
The hunt for the elusive particle began in 1964 when theoretical physicist Peter Higgs, and others, described the mechanism that would explain the origin of mass. It took many years for the theory to be accepted by the HEP community, and then useful technology was developed, on many fronts, which accelerated the process of discovery.
Of course, the Large Hadron Collider (LHC) played a pivotal role in the discovery, but the introduction of grid-enabled computing was really the key to their success.
The roots for grid go back to the Supercomputing '95 Conference in San Diego, California. At the event, a team led by Ian Foster (from Argonne National Laboratory and University of Chicago, US) demonstrated the successful execution of a number of applications running over 17 geographically distributed sites participating in the I-Way experiment. The project used middleware called I-Soft that would later, in collaboration with Carl Kesselman and his colleagues at the University of Southern California, US, become Globus Toolkit.
In the US, Globus Toolkit continues to provide homogeneity, with eXtreme Science and Engineering Discovery Environment (XSEDE), Open Science Grid (OSG), and many other projects depending on it. In Europe, several countries and domains embraced the concept, and since 1996, additional middleware varieties have been funded and developed for specific applications. But with projects hitting up against four-to-five-year funding cycles, some have fallen by the wayside. Still enough have survived that navigating the disparate middleware presents challenges, especially in regard to global collaboration, and federated e-Infrastructures have found that heterogeneity is difficult to sustain in terms of development and funding. This is probably why the number of prevailing options in Europe dropped from five in 2007, to four in 2011, among them gLite, ARC, Globus Toolkit and UNICORE – with UNICORE being the only one that does not include Globus components. Of the four, only Globus Toolkit and UNICORE are common to PRACE and EGI and have the ability to bridge the e-Infrastructures by offering a common interface to the user. In the US, OSG continues to depend heavily on both Globus Toolkit and Condor Project software as well as community-developed software for handling its massive amounts of data and jobs.
In late 2002, the HEP community formed a coordinated effort known as the Large Hadron Collider (LHC) Computing Grid, or LCG, which leveraged LCG-2 middleware. This would become their high-throughput highway to the LHC at CERN (European Organization for Nuclear Research) near Geneva, sited between Switzerland and France. LCG involved high-throughput distributed resources from the OSG in the US and Europe's Enabling Grids for E-sciencE (EGEE, which became European Grid Infrastructure, EGI, in 2010). There were four major experiments at CERN, but the ATLAS and CMS (Compact Muon Solenoid) projects were launched to cross-check and verify Higgs boson findings.
ATLAS and CMS each represent a vast multinational collaboration of more than 3,000 physicists from 41 countries and 179 institutes, with some overlap. They built upon research by many projects which leveraged the Large Electron Positron (predated LHC at CERN); the US Department of Energy's Tevatron Collider at Fermilab; and the Stanford (University-US) Linear Accelerator Center (SLAC). In 2010, high energy capability was introduced to the LHC (first operational in 2008). That's when the HEP community finally had what they needed to prove Higgs' theory on the 4th of July, 2012. EGI Deputy Director Catherine Gater chronicled the five years leading up to the discovery in an International Science Grid This Week (iSGTW) feature.
While the global HEP community was first to embrace grid technologies to this extreme, today research teams from all arenas span the globe in pursuit of life-transforming discoveries. Their workflows include a variety of resources and leverage advanced networks to engage the high-throughput systems represented by EGI and OSG, plus high-performance supercomputers (HPC), storage, visualization resources, and expertise offered by the Partnership for Advanced Computing in Europe (PRACE) and the eXtreme Science and Engineering Discovery Environment (XSEDE) in the US. To facilitate this diversity, XSEDE includes access to OSG as a supported resource allocation request. There is also a joint process that allows EU-US collaborative teams to submit unified requests for allocations of PRACE and XSEDE resources (the 2012 deadline is September 15, 2012).
Last spring's EGI Community Forum in Munich, Germany, was co-located with the Initiative for Globus in Europe's (IGE) annual user conference, and the European Globus Community Forum (EGCF). During the conference, IGE signed a memorandum of understanding with the European Middleware Initiative (EMI), a close collaboration of Europe's major middleware providers. IGE and EMI deliver middleware components for deployment by European e-Infrastructure providers that facilitate multinational collaboration. Through IGE and EMI's relationship with EGI, a quality assurance process was established to specify requirements, test, solicit feedback, and apply lessons learned in an effort to continuously improve EGI's offerings.
EMI is a three-year project that engages European users and global infrastructure providers to assess specific needs, identify redundancies, and develop a collection of consolidated and harmonious software components. Deliverables include three major releases and subsequent minor revisions, as necessary. Each set is designed to comply with open-source guidelines and to integrate with Europe's mainstream operating systems. Major releases include Kebnekaise (EMI-1, 12 May, 2011); Matterhorn (EMI-2, May 21, 2012); and Monte Bianco (EMI-3, February 28, 2013).
Although many consider the Globus Toolkit to be US software, it is open source and its developer and user communities include many Europeans who recognize its value. On October 25, 2010, IGE's roadmap was presented by Steve Crouch (UK-University of Southampton) and Helmut Heller (Germany-LRZ) at the first EGI Technical Forum in Amsterdam. At that time, EGI's Unified Middleware Distribution (UMD) officially recognized IGE as a technology provider. Their plan included timelines for the integration of resources by European e-Infrastructure providers, including EGI, PRACE, and EU-IndiaGrid2.
Globus Toolkit has been widely used in Germany since their D-Grid initiative began in 2005. The Leibniz Supercomputing Centre (LRZ) in Munich installed it on its supercomputers in 2002. Europe's fastest computer, the SuperMUC, became operational at the LRZ in August this year. SuperMUC and LRZ are committed to serve IGE-supported middleware and will most likely be driving forces for future development and use of Globus Toolkit by Europe's scientific community.
Globus Online Software-as-a-service
At the GlobusWORLD 2012 conference in Chicago last April, Foster (Globus Project co-founder) quoted the late Steve Jobs (Apple) who said "Start with the customer experience and work back toward the technology – not the other way around." Applying this philosophy and a commitment to continuous improvement, Foster and the Globus team recently launched a new effort that leverages cloud technologies to develop the Globus Online software-as-a-service (SaaS) offering. With hosted, professionally-operated services, and intuitive Web 2.0 interfaces, Globus Online aims to increase usability and functionality dramatically relative to past grid software. The SaaS model streamlines the process of delivering new features and enables the service's capabilities to be rapidly refined based on early user feedback. When EU countries add the Globus Toolkit (in particular, GridFTP and MyProxy servers) to their middleware stack, they can take advantage of Globus Online services without requiring additional software.
From left: Steve Tuecke (Globus Online, UChicago/Argonne) and IGE Program Director Helmut Heller (LRZ) at the 2011 EGI Technical Forum in Lyon, France
At the March IGE meeting, the University of Chicago's Steve Tuecke, Globus Online co-founder, presented its capabilities and anticipated future development with European interoperability in mind. Globus Online's features for high performance, secure file transfer were recently integrated with the ATLAS PanDA workload management system and it is in the testing phase. An upcoming Globus service that simplifies big-data storage and sharing could substantially enhance how the HEP community manages the massive amounts of data generated by the LHC and the new subatomic field of physics research launched by the Higgs boson discovery. Future development will target additional services to offer a comprehensive research data management solution delivered using SaaS approaches.
Of course, the biggest challenge faced by multinational collaborations is satisfying the security and privacy policies of every institution, government, and network along the way. Globus Online incorporates Globus Nexus, a service that manages user identities, including profiles, groups, and information about resources connected to the Globus research cloud. Like all Globus services, the Globus Nexus features may be accessed via a Web browser, command line, and a REST-ful programming interface that enables organizations to better integrate Globus services into their infrastructure.
The EGI Technical Forum 2012 takes place next week, from September 17-21, in Prague, Czech Republic, at the Clarion Congress Hotel. GlobusEUROPE 2012 is co-located and scheduled for Monday, September 17. The event is hosted by EGI.eu in partnership with CESNET, the consortium of Czech universities and the Czech Academy of Sciences that represents the country in the EGI Council. HPC in the Cloud is covering the event live, so check back for more coverage soon.
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?