Operating System Concepts
July 11, 2017
The strength of this book is it takes the idea of an operating system and shows it in different contexts (distributed systems for example) and pushes the boundaries of the idea itself. It gets you to think hard about what an operating system even is. Until I read this book, I only had the common but naive understanding from working with desktops and laptops.
As this is primarily and engineering book, there is a lot of technical details, so I would suggest it's best to scan and familiarize yourself with the landscape first, rather than trying to memorize everything from beginning to end.
—
I had originally studied this book to find out where operating systems leave off, and where programming languages begin. I wanted to know how to be a better programmer at lower levels. If I could think the way the computer hardware coordinates the programming language code, I could plan and optimize my code better. I think I held myself up to a standard of integrity in doing so, but I also recognize this is limited for two reasons. First, it turns out if you want to know the boundary between programming languages and operating systems, then studying assembly language is better for that.
Second, as for optimizations, I hear most modern compilers do a lot of optimizations in the backend unknown to the coder. You could write what you think is optimized code and the compiler will ignore it and optimize its own way. With that said, compilers still don't largely know how to optimize larger patterns of design, so it's still good practice to learn to think about efficient coding regardless, and learning how an operating system allocates and manages resources in particular does help with this.
Regardless of this hindsight, here is my review:
—
The first major chapters discuss Process Management, Memory Menagement, Storage Management. As a programmer both memory and storage management were pretty familiar already, especially having worked with linux for years, but I didn't know much about process management and really enjoyed its chapter. It's the sort of thing only system administrators would need to know about, but is still valuable for the rest of us, especially to feed ones curiosity.
The remaining chapters are more ad-hoc if you're looking for a unified narrative. I felt their section on Protection and Security was the weakpoint of the book as a whole, especially with the modern landscape of connected devices, criminal ransomware, and nation state hacking.
The chapter on Distributed Systems was very academically strong. As a theorist myself I especially enjoyed delving into more mathematical concepts and strategies including network topologies, distributed coordination, event ordering, mutual exclusion, concurrency, deadlock handling, election algorithms as well as ideas of agreement. Parallel processing is a very challenging and thus interesting area of research and practice.
The chapter on Special Purpose Systems didn't stand out that much for me. Mostly it had to do with optimizing an operating system for real-time as well as multimedia interactions. Obviously important, but I felt these topics weren't covered in great depth, nor were any amazing insights revealed.
One of the happier outcomes from reading this book is it clarified a lot of keywords I've seen throughout my perusals of general programming literature, such as threads, swapping, paging. Unfortunately the version of the book I read now has some outdated case studies such as windows xp. I can't speak to new editions, it's likely they've updated some of these examples, but even if they haven't, the concepts and techniques introduced are abstracted away enough from the specific softwares of the day, and overall I've found as of writing this review this book has aged pretty well.