HPC System Administration and Management - PG-DHPCSA Past Paper

ACTS, Pune Suggested Teaching Guidelines for HPC System Administration and Management – PG-DHPCSA March 2024 Duration: 60 class room hours + 60 Lab hours Objective: To introduce HPC System Administration and Management. Prerequisites...

ACTS, Pune Suggested Teaching Guidelines for HPC System Administration and Management – PG-DHPCSA March 2024 Duration: 60 class room hours + 60 Lab hours Objective: To introduce HPC System Administration and Management. Prerequisites: Knowledge of Computer Networks Evaluation method: CCEE Theory exam– 40% weightage Lab exam (Case Study based) – 40% weightage Internal exam – 20% weightage List of Books / Other training material Text Book:  High Performance Cluster Computing: Architectures & Systems (Volume-1) by Rajkumar Buyya, Pearson Reference: 1. An Introduction to Parallel Computing: Design and Analysis of Algorithms (Authors: Vipin Kumar, Ananth Grama, Anshul Gupta, George Karypis) 2. Parallel programming in C with MPI and OpenMP (Author: Michael J. Quinn) 3. Distributed Computing and Networking: 11th International Conference, ICDCN 2010, Kolkata, India, January 3-6, 2010, Proceedings (Lecture Notes in Computer... Computer Science and General Issues) 1st Edition. Edition 4. Distributed Computing Author: Seema Shah, Sunita Mahajan, Oxford Publications Note: Each session mentioned is for theory and of 2 hours duration. Lab assignments are indicatives, faculty need to assign more assignments for better practice. Data Center: Design & Management (14 Hrs Theory) Session 1 & 2 Lecture: o Data center overview o Design issues Session 3 & 4 Lecture: o HVAC o Power sizing Session 5 Lecture: o Data center matrices and best practices o Security & safety Session 6 & 7 Lecture: o Collection, rejection and reuse of heat o Liquid cooling on data centers o Energy use systems o Cabinet & cable Management Assignment: PG-DHPCSA Page 1 of 3  ACTS, Pune Suggested Teaching Guidelines for HPC System Administration and Management – PG-DHPCSA March 2024 o Case study about Data Center and Visit of Data Center Ecosystem: Architecture of HPC Cluster (30 Hrs Theory + 44 Hrs Lab) Session 8 & 9 Lecture: o Requirement Analysis Session 10 &11 Lecture: o Building blocks of HPC Session 12 & 13 Lecture: o Hardware and software selection process o Cluster Planning o Adapting Standard Linux for HPC environment (Configuration and feature selection) Session 14 &15 Lecture: o Design of HPC Cluster Session 16 &17 Lecture: o Architecture and Cluster software Session 18 &19 Lecture: o Cluster building tools Session 20 & 21 Lecture: o Multicore-architecture o Pascal o Accelerator cards o Configuring & setting environment for accelerator cards (CUDA Library) Session 22 Lecture: o Latest trends and technologies in HPC o Case study: Param Shavak and Use Cases of Param Shavak for HPC solutions Assignment: o Write Survey Paper on Multicore processor and latest advancement in this HPC System Management and Monitoring (16 Hrs Theory + 36 Lab Hrs) Session 23 Lecture: s o IPMI o HMC Session 24 & 25 PG-DHPCSA Page 2 of 3  ACTS, Pune Suggested Teaching Guidelines for HPC System Administration and Management – PG-DHPCSA March 2024 Lecture: o User management using LDAP/NIS o Processor usage, memory usage o Network monitoring, network usage o Gangila, Nagios o Node resources Session 26, 27, 28, 29 & 30 Lecture: o System Benchmarking o Theoretical peak performance o HPL bench mark, Tuning HPL, Problem size, Block size, process grid PxQ Assignment: o operate, maintain, integrate, upgrade, and manage all HPC resources including related hardware and software Assignments –Lab: o Data Centre visit o Building a manual HPC Cluster o Building an HPC Cluster using different Cluster building and management tools o Monitoring tools installation & configuration o Network monitoring using Nagios o IPMI configuration o System benchmarking using HPL o Case study HPC Solution (PARAM Shavak) PG-DHPCSA Page 3 of 3