The AVES computing system, based on an "Cluster" architecture is a fully integrated, low cost computing facility dedicated to the archiving and analysis of INTEGRAL data. AVES is a modular system managed by a software resource manager (SLURM) that allows unlimited expandability (65,536 nodes and hundreds of thousands of processors); actually is composed by 30 Personal Computers with Quad-Cores CPU able to reach the computing power of 300 Giga Flops (300x109 FLoating point Operations Per Second), with 120 GB of RAM and 7.5 Tera Bytes (TB) of storage memory in MHDDFS configuration. AVES was designed and built to solve growing problems raised from the analysis of the big data amount accumulated by INTEGRAL mission (actually about 8 TB) and due to increase every year. To this purpose we have developed a software interface able to split the required analysis process in N processes automatically sent to N cores. Thus the whole computing time, compared to that needed by a Personal Computer with single processor, has been enhanced up to a factor 70.
AVES: A computer Cluster System approach for INTEGRAL Scientific Analysis
Martino B;
2009
Abstract
The AVES computing system, based on an "Cluster" architecture is a fully integrated, low cost computing facility dedicated to the archiving and analysis of INTEGRAL data. AVES is a modular system managed by a software resource manager (SLURM) that allows unlimited expandability (65,536 nodes and hundreds of thousands of processors); actually is composed by 30 Personal Computers with Quad-Cores CPU able to reach the computing power of 300 Giga Flops (300x109 FLoating point Operations Per Second), with 120 GB of RAM and 7.5 Tera Bytes (TB) of storage memory in MHDDFS configuration. AVES was designed and built to solve growing problems raised from the analysis of the big data amount accumulated by INTEGRAL mission (actually about 8 TB) and due to increase every year. To this purpose we have developed a software interface able to split the required analysis process in N processes automatically sent to N cores. Thus the whole computing time, compared to that needed by a Personal Computer with single processor, has been enhanced up to a factor 70.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
