Grid computing – Understanding various types of grid systems



Currently there are three main types of grid systems :

Voluntary grids – grids through the use of voluntarily provided free resource of personal computers;
Scientific grids – application largely programmed in a special way (eg using Globus Toolkit);
Grids based on the allocation of computing resources on demand (commercial grid, Enterprise grid) – ordinary commercial applications run on a virtual machine, which in turn, consists of several physical computers together using Grid technologies.

The term ” grid computing” appeared in the early 1990s as a metaphor, showing the possibility of easy access to computing resources as well as to the mains (power grid).

Grid technology is used for simulation and data processing in experiments at the Large Hadron Collider (grid used in other computationally intensive tasks).

BOINC platform is currently under active computing with more than 60 projects. For example, the project Fusion (south of France, to develop a method to generate electricity using nuclear fusion experimental reactor ITER also uses a grid (EDGeS @ Home).

Titled CLOUD, the project started commercialization of Grid technology, in which small companies, institutions, requiring computational resources, but can not afford te cost, can purchase grid computing time at the Supercomputer Center.

Structure Grid Systems CERN

CERN Grid System, designed for processing of data from the Large Hadron Collider , has a hierarchical structure.

Highest point of the hierarchy, zero – CERN (getting information from the detector, the collection of raw scientific data that will be stored until the end of the experiment). During the first year of operation, it is planned to collect up to 15 petabytes (thousands of terabytes) of data of the first copy.

The first level , Tier1 – deposit of the second copy of the data in other parts of the world (11 centers : in Italy, France, UK, USA, Taiwan, and a center of the first level – CMS Tier1 – CERN. Centers have substantial resources for data storage.

Tier2 – following in the hierarchy, many second-tier centers. Having large storage resources is not necessary; but must possess good computational resources.

Russian venues in Dubna (JINR) , three centers in Moscow ( SINP , LPI , ITEP) , Troitsk (INR ) , Protvino (IHEP ) , St. Petersburg ( St. Petersburg State University ) and Gatchina ( PINP ) .

In addition, a network associated with these centers and other centers of the JINR Member States – in Kharkov, Minsk , Yerevan , Sofia , Baku and Tbilisi.

More than 85 % of all computing needs of the LHC as of 2010 was carried out at CERN , of which more than 50% at secondary care centers.

About the Author