Brazil stood out before the international scientific community with the entrance of the High Energy Experimental Physics Laboratory (Lafex), of the Brazilian Center for Physics Research (CBPF), of the Relativistic Heavy Ions Group (Griper) and of the Physics Institute of the University of São Paulo (USP) into the exclusive team of computer grids that take part in the LHCONE network, managed by the European Organization for Nuclear Research (Cern).

In operation since 2011, LHCONE is a network infrastructure service conceived exclusively for the projects of the largest particle accelerator in the world, the Large Hadron Collider (LHC). In it, experiments take place, which generate such a large quantity of data that extrapolates the resources of Cern’s own data center, with 100 petabytes, equivalent to 700 years of Full HD quality films. Due to that, the processing of such data is distributed among 150 computer centers spread out in 40 countries, which form the largest computer grid in the world, the Worldwide LHC Computing Grid (WLCG).

In order to ensure greater safety and priority for the data, in addition to a better network outflow and performance, LHCONE works as a network juxtaposed to the existing physical network, interconnecting several websites of institutions that house computer grids that are LHC’s collaborators, by means of the academic networks. That is the case of Lafex, which in July 2015 started to fulfill the minimum network requisites required by Cern to be a member of the exclusive network.

However, for it to happen, it was necessary to overcome certain performance issues, given that the CBPF is connected to the Brazilian academic network in an indirect manner, through the Rio metropolitan network (Rede Rio), with a band capacity that evolved until it reached the current 10 Gbps.

Since 2010, the Brazilian National Research and Educational Network (RNP) focused on mapping out the data transfers made by Lafex, in order to find possible flaws in the pathway between the laboratory and the national backbone, and then between the international connections and Europe. “We used monitoring platform perfSONAR, which helped us to visualize where the issues where occurring along such path”, says RNP’s Experimentation Networks Manager, Alex Moura.

In addition to RNP, Rede-Rio/Faperj and teams from large international backbones of the Latin American Cooperation of Advanced Networks (RedCLARA), from Géant and from Cern contributed to the success achieved. A part of the project received funding from the Institutional Qualification Program (PCI) of the Ministry of Science, Technology and Innovation (MCTI).

Computer Grids in Latin America

The CBPF collaborates with Cern for project LHCb (Large Hadron Collider beauty), which investigates the differences between matter as it is known and antimatter, by studying a kind of particle called B-quark. Said particles are flung after the collision in the accelerator and captured by a series of subdetectors present in the LHC. The project relies on the collaboration from about 700 scientists from 66 different institutions and universities.

The computer grid with the greatest data processing capacity in Latin America is Griper’s, from the Physics Institute of the University of São Paulo, which receives up to 75% of the entire processing of the LHC in the region. Recently incorporated into the LHCONE network, Griper takes part in the LHCb and Alice projects. In the latter, the collisions of heavy ions, such as lead, are measured to study a new state of matter, the so-called Quark-Gluon Plasma. Made up of the most basic elements of the matter, it may explain complex physics issues, such as the origin of the universe.

In August 2015, Griper was also responsible for the upgrade of equipment of the LHC for the processing of data from Alice. The chip, named Sampa and developed in partnership with the polytechnic school of the University of São Paulo, is one of the items of equipment that will record in images the exact moment of said collisions.

The third Brazilian computer grid connected to the LHC is that of the São Paulo Analysis and Research Center (Sprace), from the Universidade Estadual Paulista (Unesp). Sprace contributes with the CMS (Compact Muon Solenoid) project, which is a detector that uses a large magnet to identify the path of the particles produced by the collisions in the accelerator. This project was essential for the discovery of the Higgs Boson subatomic particle and the composition of the dark matter, which redirected the current studies in Astronomy.

Another computer grid that processes data from the CMS project is located in the High Energy Physics Laboratory (HEPGrid) of the Rio de Janeiro State University (Uerj). In December 2014, with support from RNP, the HEPGrid took part in event Supercomputing, where a world record in data transmission was broken. The group achieved a transmission rate of approximately 1.4 terabits per second, sustained for a prolonged period. The maximum rate ever obtained in Supercomputing was 750 Gbps. The success was credited to the use of the SDN technology, of networks defined by software.

According to the February 2014 data, the CMS is one of the largest international scientific experiments in history, due to the fact that it involves 4.3 thousand professionals, including particle physicists, engineers, technicians and students from 182 research institutions from 42 countries.

The LHC

The LHC was built between France and Switzerland and has a tunnel with a circumference of 27 kilometers with electromagnetic superconductors, where the particle collision takes place. It is managed by Cern, created after the end of World War II. The LHC currently engages in four experiments: Alice, Atlas, LHCb and CMS. Together, they involve over 8 thousand researchers from the world over, including from Brazilian research centers.

The computer grid that serves the LHC is divided into four levels or layers, the tiers, from 0 to 3. T0 is Cern’s datacenter, through which all data produced by the LHC pass in raw form until they reach T1. The latter consists of 11 computer centers, in the United States, Europe and China, with sufficient size to store all data generated by the accelerator and which are responsible for reprocessing them in large scale and distributing them to those at level T2, in charge of more specific tasks. Brazil has four T2-level clusters: Lafex, from the CBPF, Griper, from USP, Sprace, from Unesp, and HEPGrid, from the State University of Rio de Janeiro (Uerj).