Supercomputers for Linux SysAdmins Managing Modern HPC Clusters and Supercomputers from Software to Hardware (Sergey Zhumatiy) (Z-Library)

【One-Line Pitch】 A practical bridge from Linux system administration to running real HPC clusters and supercomputers, covering the hardware, interconnect, and operational mindset that separates a cluster from a single server. Best for experienced Linux admins moving into HPC, with enough grounding for motivated newcomers. 【Book Arc】 - **Opening (~0%–13%)**: Frames the book's purpose and audience, sets conventions and notation, and positions it as a companion to hands-on Linux practice rather than a standalone tutorial. - **Early (~13%–27%)**: Defines what makes a machine "super" — parallel processing concepts, cluster types, and how clusters differ from supercomputers — then moves into building and starting a system: anatomy, planning, and documentation. - **Middle (~27%–60%)**: The excerpt coverage here is thin; the table of contents suggests this stretch continues the build-and-operate material, but the excerpts do not detail its contents. - **Late (~60%–93%)**: Hardware and interconnect come into focus: control, compute, login, and service nodes; network equipment; data storage; hardware architecture features; and a dedicated treatment of InfiniBand, including addressing, subnet management, IPoIB, and management utilities. - **Ending (~93%–100%)**: Shifts to operations and fundamentals — how a typical user session and job life cycle work, what stays hidden from users, and a refresher on UNIX/Linux basics such as processes, access rights, services, and manuals. 【Key Takeaways】 - **This book targets a specific career transition** (Opening): it assumes Linux admin experience and aims to make you a supercomputer administrator, not to teach Linux from scratch. (Opening) - **"Super" is an operational concept, not just a speed label** (Early): the book distinguishes clusters from supercomputers and explains what "super" means specifically to the administrator, including centralized management of the whole complex. (Early) - **Planning and documentation precede hardware** (Early): before procurement or assembly, the book stresses planning and documentation as first-class steps in building a system. (Early) - **Node roles are the organizing principle of HPC hardware** (Late): control, compute, login, and service nodes each carry distinct responsibilities, alongside network equipment and data storage. (Late) - **InfiniBand gets dedicated, hands-on coverage** (Late): component identification and addressing, subnet management, IP over InfiniBand, and utilities for viewing and managing the fabric, plus alternatives. (Late) - **Job life cycle is the admin's mental model** (Ending): understanding how a user session unfolds and what is hidden from the user is presented as core operational knowledge. (Ending) - **Linux fundamentals still matter at HPC scale** (Ending): processes, access rights, key services, and manuals are revisited because they underpin everything above them. (Ending) 【Reading Tips】 - If you are already a seasoned Linux admin, skim Chapter 1 as the author suggests and spend your attention on the HPC-specific chapters (hardware, InfiniBand, job life cycle). - Treat the planning and documentation sections as checklists for real deployments rather than theory — they are easy to skip and costly to ignore. - Pair the book with hands-on practice, ideally on a virtual machine or test cluster, and revisit chapters after you have operational experience. - Use the "Brief Summary" and "Search Keywords" sections at the end of each chapter as quick review anchors when returning to the book later. 【Coverage Limits】 The excerpts are heavily front-loaded with front matter and table-of-contents entries; the middle portion of the book is largely uncovered, so this guide cannot describe its specific content beyond what the chapter listing implies.