PIC Scientific Information Port (Port d’Informació Científica)

The PIC data centers are located in UAB building D, with 24×7 access control. The technical rooms where the PIC computer equipment is housed have high-availability power supplies, including a 500 KVA UPS and diesel generators that cover the building's total demand, as well as a high-efficiency cooling system. The PIC has two different data centers in the same building: ● 150 m2 air-cooled room that houses the computer storage and calculation equipment. It includes an innovative hot air containment system above the ceiling that increases efficiency. The hot air is cooled by three indirect freecooling units supported by an adiabatic installation. The room has a fire detection and protection system. ● 25 m2 room with liquid immersion cooled systems that only houses computing resources. It has four Green Revolution Cooling tanks, each of which can contain 45U of computer equipment and can dissipate up to 45kW of heat. Compute cluster: 160 compute nodes that provide a total of: ● 12000 CPU cores ● 18 GPUs ● 34 TB of RAM ● 524TB of local disk Disk storage system: ● Managed by dCache software ● Installed capacity of 20 Petabytes ● Access from compute nodes ● Remote access via web and with protocols https/webdav, gridftp, xrootd, scp Magnetic tape storage system: ● Managed by Enstore/dCache software ● Installed capacity of 63 Petabytes ● Access from the computing nodes ● Remote access with protocols https/webdav, gridftp, xrootd, scp Network: ● 200Gbps connectivity to the Anella Científica ● Internal network 100Gbps ethernet---

1. Data Management ● Multi-Petabyte scale data storage while keeping it accessible with access control. ○ Scalable and high-performance data space for analyzing and sharing data via the web. ○ Long-term data preservation. ● Consulting to design, deploy and operate automated data processing and analysis workflows. 2. Data Analysis ● Cluster with CPUs and GPUs to execute data analysis jobs. ● Web access to CPU and GPU resources using jupyter notebooks. ● Consulting to deploy analysis workflows and accelerate performance. 3. Software Development ● Platform Gitlab website to manage version control and CI/CD. ● Advice on implementing DevOps methodologies in scientific software development. 4. Training ● Python ● Machine Learning

GINYS-IFAE-004

Obert