Computing

CESGA has different computing platforms of different architectures to allow the researcher to always choose the architecture that best suits their calculation needs.
For simulations that require calculation of high performance and supercomputing, the FinisTerrae supercomputer offers higher performance and a high performance interconnection network for parallel work or that require the use of GPUs. It also allows simulations that require handling large volumes of data.
For application virtualization, simulations on demand or that require interacting with the user, the cloud platform offers these features and the flexibility to adapt to the requirements of different operating systems or hardware configurations.
The BigData environment, meanwhile, is optimized for the massive processing of data using the latest technologies and software oriented to data management and processing.

FinisTerrae II

FinisTerrae is the generic name of the different generations of supercomputers of the Supercomputing Center of Galicia.

FinisTerrae is an advanced computer equipment, integrated in the Singular Technical Scientific Facility (ICTS), Spanish Supercomputing Network (RES).

These teams have been partially financed with funds contributed by the Xunta de Galicia, the Higher Council for Scientific Research, the European Regional Development Fund (ERDF) and the Ministry of Science, Innovation and Universities.

In 2007, the first equipment of the FinisTerrae series was installed, which then reached the 100th position in the list of the 500 most powerful computers in the world and recognized as the Singular Technical Scientific Installation of Spain.

The equipment currently in service, the FinisTerrae-II was installed in 2015 and put into production in 2016. It is a Bull ATOS bullx equipment distributed in 8 racks or cabinets and has 320 computer nodes, 7,712 cores, 44,544 GB of memory and 750,000 GB of high performance storage Luster. All processing and computing nodes are interconnected through a Mellanox Infiniband FDR low latency network. The peak computing capacity of the equipment is 328,272 Gflops and the sustained performance obtained in the Linpack test is 213,000 Gflops.

In 2017, the Galician Virtual Supercomputer (SVG) nodes used for distributed calculation tasks were integrated into the FinisTerrae-II queue system to provide a single point of access to the entire computing infrastructure of the center.

Access

There are different mechanisms to access the equipment. There is an account request form for the research groups of the Galician university system and the Spanish council of scientific research, CSIC, (https://www.altausuarios.cesga.es/). Research groups that do not belong to this population may request access through the Spanish Supercomputing Network (https://www.res.es/acceso-a-la-res), the European PRACE supercomputing network (in DECI calls http://www.prace-ri.eu/deci-projects/). Cesga promotes the use and facilitates access to this equipment to R+D+I departments from companies and institutions (contact:  sistemas[AT]cesga.es).


Use cases

Since its installation, the team provides advanced computing services to more than 500 research, development and innovation groups in Galician centers and universities, of the Higher Council of Scientific Research, administrations and companies, in addition to providing calculation capacity to the Spanish Supercomputing Network (RES) and the European supercomputing initiative Partnership for Advanced Computing in Europe (PRACE) through the calls Distributed European Computing Initiative (DECI). Each year it provides more than 70 million hours of computation to different scientific and knowledge areas.

The Cloud infrastructure is composed of a set of servers and storage systems in which it is possible to provide users with a virtual computing infrastructure that can be customized to the end user requirements: operating system, number of processors, memory, disk and Number of nodes are determined to suit the user dynamically.
OpenNebula software is used for system management and a graphical interface is available to simplify its use to end users, in addition to a command line interface for more advanced users.

Cableado cloud computing

The Big Data infrastructure allows the processing of large volumes of information in parallel, using the power of modern tools that greatly simplify the process.

The service has two different environments to meet the needs of all users:

  • Environment CDH 6.1.1 (Hadoop 3): Offers a platform based on Hadoop 3 that supports the main components of the Hadoop ecosystem.
  • Environment Platform as a Service (PaaS): For cases where you need to go beyond what the Hadoop ecosystem offers: Cassandra, MongoDB, Redis, …

The infrastructure consists of a dedicated cluster of 38 nodes (4 master nodes and 34 slave nodes). It has a gross capacity of 816 TB and an aggregate I / O capacity of more than 30 GB / s and has a 10 GbE connection between all cluster nodes.

The platform has a web interface (WebUI) to simplify the use of the platform to new users who have not had previous contact with Big Data tools. The WebUI allows you to perform the most common tasks in a simple way and is a complement to the command line.