Scientists who found God's particles are considering turning to cloud computing

The discovery of the Higgs boson, the "God particle", is the most revolutionary physics discovery in recent decades, so how do European scientists do it? From an IT perspective, they rely on an excellent, somewhat old-fashioned grid computing http://www.aliyun.com/zixun/aggregation/13748.html "> Infrastructure, although a new, The cloud based computing infrastructure has already appeared on the horizon.

The European Organization for Nuclear Research (CERN), which has been deploying the grid computing infrastructure for 10 years, has been widely used by researchers over the past few years to help discover the Higgs boson, or the so-called "God particle", to make it famous overnight.

CERN's grid is very different from the public cloud, which hosts data and resources in one or more centrally managed data centers, where users connect to these resources, while CERN's grid connects more than 150 computing sites around the world, sharing information with each other.

At the time the CERN grid was just built, it processed data at 15PB to 20PB a year. This year, CERN has handled more than 30PB of data. "If all data processing is done by CERN itself, it's definitely not going to work," said Ian Bird, head of computing grid project at CERN. "Grid computing is actually very similar to today's cloud computing. "In a way, we're already in the clouds," he said.

The Large Hadron Collider (LHC), found in the Higgs boson, is located at CERN's workplace, the center of the entire research activity, and its grid uses a tier 0 connection. In the LHC's 17-mile tunnel, the various experimental data generated by the strong collisions of the particles will be sent through the grid to 11 Tier 1 sites, which are renowned laboratories with large data centers that can handle massive amounts of scientific data. The data sets generated by these sites are distributed to more than 120 academic institutions worldwide for further testing and research.

The entire grid has a disk capacity of up to 200PB, with 300,000 computing chips connected to more than 150 computing centers via a 10Gbps link. "The grid connects all of this together to make it look like a single system." "The vast majority of sites are equipped with standard Red Hat Linux distributions, as well as custom storage and computing interfaces, and each site provides information services to depict data."

Of course, research that contributes to landmark discoveries such as the Higgs boson is not always centrally managed. In fact, this is a fairly confusing process and makes it difficult for them to develop a plan for allocating the appropriate amount of resources to the test work for each site, Bird said. For example, after a particle collision within the LHC, the affected particles vanish instantaneously in the detector. The first phase of data analysis is the reconstruction of the collision process, followed by tracking the trajectory of various particles, mainly in the tier 0 (i.e. CERN) and Tier 1 sites. Other stages of analysis are broken down into smaller datasets and distributed to collaborative academic institutions for analysis. As a result, a large number of statistical analysis, histogram analysis and data mining will follow. If a discovery is to be made, the analysis should be very precise and can be used for repeated tests.

That's why bird and CERN are quite excited about the potential of cloud computing. "We are interested in using cloud storage," he said. "But at this point, we can't determine the cost of using the cloud and how it will affect our funding structure." "The main source of funding for CERN is the data generated by various academic institutions visiting CERN for analysis and paying a fee for this." Many of these cooperative academic groups have their own computing resources and want to experiment with CERN's data on their own sites, allowing them to be used in their academic communities. "From a technical point of view, the cloud model can be run, but I don't know how to charge." ”

CERN has taken the first steps to cross the cloud. Internally, CERN is running a private cloud based on OpenStack Open source code. Many cooperative organizations also build private clouds on their own sites.

This March, CERN and two other European research institutions also built a public cloud called Helix Nebula, also known as the Science cloud. This is the result of collaboration between research institutions, cloud vendors and IT support companies, and can be viewed as a community cloud for the research community. CERN has recently launched a two-year pilot program to simulate the LHC experiment in the Helix Nebula Cloud.

Bird is full of hope for cloud computing, that the next 10 years, the cloud is the current grid. "It is not clear how this can be achieved." "But even if cloud computing is full of challenges, Bird is convinced that scientists who can do great scientific discoveries such as God's particles will certainly be able to understand and master cloud computing." (Wave Compilation)

(Responsible editor: The good of the Legacy)