The infrastructures of the JIC project

High performance computing cluster with 2000 cores, Intel E5 2680 V3 with more 14 TB RAM

Big Data cluster with 576 Cores, Intel E5 2683 V3 and Intel E5 2618L V3 with more than 12 TB RAM

Huawei OceanStor 6800V3 Storage System with a capacity of 1.1 PB on different disk types (SSD, SAS, NearLine SAS)

Storage system for BigData Huawei OceanStor 5300 V3 with about 100TB on SAS disks

Network security management equipment - firewall (with throughput up to 80Gbps in cleartext)

Video Surveillance Management System (Video Content Management) with 20 TB SAS storage system for image storage and Video Cloud node

Geographic Information System (GIS)

Core Network network equipment for LAN management (very high performance laboratory local network) with a total of 144 1 / 10Gbps and 100Mbps RJ45 ports, 480 10Gbps SFP + ports, 84 40Gbps ports

Complete system for the creation and implementation of a private LTE network (e-LTE) including CoreNetwork, antennas, base station and management system

Experimental e-LTE Rapid Solution system for field trials in case of emergencies / disasters

IOT Gateway for managing the data flows of the various sensors

Antenne e-LTE

Huawei e-LTE chipset for sensors

IOT Gateway for managing the data flows of the various sensors

The sensors are mainly of two types:

With Huawei Chipset, directly connected with e-LTE technology

OPEN, available on the market linked to the 'Metropolitan Digital Fabric' project

We proceeded with the installation, configuration and management of the following software:

Intel Compiler, PGI compiler, GNU  compilers, openmpi,  Mvapich, Mpi2

All software (moloch etc) for their weather forecast pipeline, libraries and python integrations

Openstack that allows you to instantiate virtual machines with the operating system and software you want

Docker container system. User freedom on Docker containers is practically total, users can load and use software or operating systems available in the form of containers

On fat-node machines, those with GPUs, in addition to what has already been mentioned, there are several tools for using the GPU, Nvidia-docker and others

Fluid dynamics and hydrodynamics side have been installed: starccm, xfoil, dakota, paraview, salome, altair HyperWorks

On the docker side, intense work has been done to configure the Kubernetes container management environment within the SGE system. The integration of Kubernetes with SGE (Sun Grid Engine) is complex and laborious, the ultimate goal is to instantiate a Kubernetes environment on the entire cluster, with a simple SGE command. The added value of this solution is that Kubernetes can potentially extend to the entire cluster, if this is available, as a normal job, in competition with all other user jobs and with more traditional usage needs that already operate on the cluster