In the beginning of February, the first computer cluster of the University of Tartu was opened in the Scientific Computation Centre of the university; this is one of the most powerful clusters in Estonia and consists of 42 computation servers, a control device and a storage device.

The implementation of this cluster provides good opportunities for developing such scientific fields that need processing large volumes of data or that use computer power for modelling various processes, e.g. gene technology, bioinformatics and material sciences.

The opening of the cluster is an important milestone for the Scientific Computation Centre of the University of Tartu. The Centre was established in the beginning of summer 2008. The goal stated by Jaak Vilo, one of the main initiators of the Scientific Computation Centre, i.e. “to get a large cluster for use by scientists, with many computers, each more powerful than usual desktop machines”, is now fulfilled as the new cluster is opened.

The best tender conforming to the procurement requirements was made by MicroLink Eesti. According to Üllar Feldschmidt, Sun Microsystems Category Manager of MicroLink, co-operation between higher education institutions and scientific institutions has always been very important for MicroLink and Sun Microsystems. The University of Tartu also has a long-term relationship with MicroLink and an important part of this has been the use of Sun technology in scientific and educational work.

“The opened cluster is currently one of the most powerful in the Baltic States and also one of the most modern clusters regarding the technology used. The current solution allows for further extension of the system on a very wide scale. The long-term experience and innovation of Sun Microsystems in constructing clusters with high computation capacity, and the open software and the energy-efficient hardware design has allowed us to achieve low expenses for acquiring and operating the cluster as well,” said Feldschmidt.

• The cluster consists of a control computer, a 9 TB storage device and 42 computation servers;
• Every server has 8 processor cores, 32 GB or RAM and a 500 GB hard drive;
• All computers are connected to each other with a fast (20 Gb/s) InfiniBand connection having low latency; this allows successful use of programs needing intensive parallel communication;
• Intel Xeon E5400 series processors with low current consumption are used;
• The maximum theoretical computation capacity of the cluster is 840 GFlops (840 billion flowing point operations per second).