Arquitectura clúster de alto rendimiento utilizando herramientas de software libre.
In recent years science and technology have made great advances with respect to the high performance computing, framed in architectures specifically designed for their operation and characterized by its high cost, being unreachable by research centers and higher education institutions (IES). Computi...
Furkejuvvon:
| Váldodahkki: | |
|---|---|
| Materiálatiipa: | bachelorThesis |
| Giella: | spa |
| Almmustuhtton: |
2015
|
| Fáttát: | |
| Liŋkkat: | http://dspace.unl.edu.ec/jspui/handle/123456789/11025 |
| Fáddágilkorat: |
Lasit fáddágilkoriid
Eai fáddágilkorat, Lasit vuosttaš fáddágilkora!
|
Lasit fáddágilkoriid | Čoahkkáigeassu: | In recent years science and technology have made great advances with respect to the high performance computing, framed in architectures specifically designed for their operation and characterized by its high cost, being unreachable by research centers and higher education institutions (IES). Computing for researches use powerful tools and methods to process data in advanced academic researches. With a cluster built by conventional hardware and free software tools, the Systems Engineering Degree of National University Loja can obtain architecture with the speed and power of an expensive supercomputer at a fraction of the minimum cost. The purpose of this project is to provide and encourage academic and scientific research, teachers and students from the CIS, obtaining good results in: simulation process, neural network learning, pictures processing, patterns recognition, database processing, processing of large volumes of Big Data information, data mining, decryption of codes, assessment of algorithms, etc. The implementation of a platform of higher yield allows made billions of operations per second, allowing significant advances into the academic inquiry in any field of research into the university area. The implementation of the cluster was done with the GNU / Linux Debian "Wheezy" operating system, by its stability and efficiency as a server of high-performance used like a standard for the development of this project. For middleware we used Mosix in version 3.4.0.12 like a manager of process to get a unique image of the system oriented to distribute computing. MPICH was used in version 3.1.2 for parallel programming as one of the standard applications deployed Message Passing Interface (MPI). For the rendering process we used Blender, for the analysis of Big Data we used the Apache Hadoop framework. Among the configurations that we made they were: Remote access using public keys through Secure Shell protocol (SSH), sharing of network system through the Network File System protocol (NFS), also the design of network topology and the configuration of the motorization system Ganglia Monitoring System. |
|---|