for College Students
Phase 0 — Right Now, Before Job Starts (Next 2–3 months)
Section titled “Phase 0 — Right Now, Before Job Starts (Next 2–3 months)”This is your runway. Use it well but don’t burn out.
Subscribe to newsletters — yes, do this first, takes 10 minutes total. SRE Weekly, ByteByteGo, Cloudflare Blog RSS. Don’t read everything, skim headlines. You’re building pattern recognition, not knowledge yet.
One book at a time, in this order:
Start with How Linux Works by Brian Ward. You said you have basic Linux knowledge — this book will fill the gaps you don’t know you have. Boot process, kernel, filesystems, networking from the OS perspective. It reads fast, ~2-3 weeks. Don’t move on until you finish it.
Then OSTEP — https://pages.cs.wisc.edu/~remzi/OSTEP/ — free online. Don’t read every chapter. Read: Virtualization (processes, CPU scheduling, memory/paging), Concurrency (locks, condition variables), Persistence (filesystems, journaling). Skip the homework unless something interests you. This reinforces what How Linux Works introduced but explains the why. Another 2-3 weeks at a casual pace.
Parallel to the above — not a book, just a daily habit: pick one Cloudflare blog post and one Julia Evans post per week. Read them like articles, not studying. The goal is exposure.
That’s it for Phase 0. Resist the urge to start DDIA or Stevens TCP/IP yet. You’re not ready to absorb them without context.
Phase 1 — First 3 Months on the Job
Section titled “Phase 1 — First 3 Months on the Job”The job changes everything. You will hit real problems and suddenly need specific knowledge. Let the job drive what you read.
Keep reading — start The Linux Programming Interface now. Don’t read it cover to cover, use it as a reference. When you encounter something at work — signals, sockets, processes — look it up in TLPI. This is how engineers actually use this book.
Start TCP/IP Illustrated Vol. 1 by Stevens. But read it slowly, one chapter a week maximum. You will encounter networking issues at the bank — DNS resolution, load balancer behavior, connection timeouts. When that happens, stop and read the relevant Stevens chapter immediately. It will hit differently when you have a real problem to map it to.
LinkedIn School of SRE → https://linkedin.github.io/school-of-sre/ — Read this in parallel with your first few months. It’s structured like a curriculum specifically for someone transitioning into SRE from a dev background. Chapter by chapter, one section per week.
Phase 2 — Months 4–9
Section titled “Phase 2 — Months 4–9”By now you have context. You’ve debugged real things, seen real infrastructure, have opinions about what confuses you.
DDIA — Start now. Read one chapter per week, no faster. This book rewards slowness. The chapters on replication, partitioning, and transactions will suddenly make sense because you’ve seen the systems they describe. Don’t rush this one.
Brendan Gregg — Start watching his talks on YouTube now. Don’t buy Systems Performance yet. Watch the LISA 2019 talk first, then “Linux Performance Analysis in 60s.” When you start debugging actual performance issues at work, buy the book.
ByteByteGo — Watch the YouTube channel regularly. These are 10-minute system design explainers. Good for HLD vocabulary building.
Phase 3 — Year 1 to 2
Section titled “Phase 3 — Year 1 to 2”Google SRE Book — selective chapters, not front to back. Read the chapters on SLOs/SLAs, error budgets, toil, on-call, and postmortems. These are the chapters that change how you think, not just what you know.
Release It! by Michael Nygard — Read this when you’ve seen your first production incident. It will feel like the author was watching over your shoulder.
Systems Performance 2nd Ed by Brendan Gregg — Now you’re ready for this. You’ve seen performance issues. You know what a flame graph is. This book makes sense now.
Database Internals by Alex Petrov — Only after finishing DDIA. This goes one level deeper.
Phase 4 — Year 2 and Beyond
Section titled “Phase 4 — Year 2 and Beyond”Linux Kernel Development by Robert Love — When you’re actually curious about the kernel, not before. This is a book you read because you want to understand scheduling or memory management at a deeper level, not because a roadmap told you to.
MIT 6.824 papers — Pick one paper per month from the syllabus. Raft, Spanner, MapReduce. Read them slowly. Discuss them with someone if you can.
The Algorithm Design Manual — Dip in when you encounter algorithmic problems. Consistent hashing when you’re working on load balancing. Bloom filters when reading about distributed caches.
FinOps / Cost Optimization — Only start this when you have actual cloud spend responsibility. Reading about it before you have bills to look at is abstract and mostly doesn’t stick.
The One Rule That Overrides Everything
Section titled “The One Rule That Overrides Everything”When the job gives you a real problem, drop the scheduled reading and go deep on that topic immediately. A book chapter read to solve a real problem is worth 10 chapters read on a schedule. The roadmap above is a default — your actual work is a better teacher.
You have maybe 2–3 focused hours per week outside work for this. Protect those hours but don’t try to fill every minute. The goal isn’t to finish books, it’s to build a mental model that compounds over time.
Week 1 action: Subscribe to newsletters, order How Linux Works, bookmark OSTEP. That’s it. Start Monday.