Cover of Fly.io Made Crystal Clear
← Back to the series

Fly.io

Made Crystal Clear

By Christophe Paka · Updated September 4, 2026

A practical, visual guide to running real apps close to your users, and knowing what it costs

Prefer it on your Kindle, or on paper? Same book, on Amazon.

Give feedback

Who this book is for

In plain words: Fly.io runs your app on servers in many cities at once, so it is fast for users everywhere, and you pay for what you use.

  • Developers running a real app close to their users
  • Teams leaving Heroku or weighing a plain VPS
  • Not for: Not for you if you only need a static website. Try Vercel Made Crystal Clear.

Fly.io runs your app as a real long-running process in a microVM, close to your users, on one anycast address. It is genuinely good at that. It also has no free tier, bills by the second, and carries a reliability history people still argue about using numbers from 2023. This guide handles all three honestly: the exact rate card, a dated outage timeline, and the volume model that has cost people data. Then the platform properly, ending with a money chapter whose answer is not always Fly.io.

Pages
154
Level
Beginner to advanced
Updated
September 4, 2026

12 real pages. Click a page to enlarge it.

Page 8: three shapes of compute compared
Page 17: the full path from local code to a served request
Page 18: the Machine lifecycle
Page 33: monthly cost if a Machine ran 24/7, by size and family, Amsterdam region, captured
Page 49: fly.toml mapped onto the platform
Page 53: an inbound request on 443 through the handler chain to your process's port
Page 59: the Machine lifecycle and what each transition costs you
Page 76: the two networks, public anycast and private 6PN, and who can reach what
Page 91: managed and unmanaged Postgres side by side, on support, HA, cost basis, and who is on
Page 94: what you need leading to MPG, a partner service, or an external database, with
Page 97: the deploy pipeline from your code to live traffic, with the health check as the gate
Page 121: portability audit

Inside the book

  1. 01What Fly.io Actually Is
  2. 02The Mental Model: Deploy to Served Request
  3. 03Your First Deploy
  4. 04What Fly.io Actually Costs
  5. 05The Reliability Question, Dated Honestly
  6. 06fly.toml and Configuration
  7. 07Machines, Autostart, and Scale to Zero
  8. 08Regions, Anycast, and Being Close to Users
  9. 09Networking: Public, Private, and Flycast
  10. 10Volumes and Persistent State
  11. 11Databases: Managed Postgres and the Legacy Story
  12. 12Deploys, Health Checks, and Rollback
  13. 13Observability and Debugging a Live App
  14. 14The Business Case, and the Alternatives
  15. 15The Playbook, Cheat Sheet, and Where to Go Next

You'll leave with

  • Tell a Fly Machine apart from a container and from serverless
  • Trace a request from anycast edge to the nearest healthy machine
  • Price your app exactly: there is no free tier, and stopped is not free
Christophe Paka, author of Fly.io Made Crystal Clear

About the author

Christophe Paka writes every book on the shelf. He spent years in technical support at Salesforce and Microsoft, where the job was to make complex systems clear to customers who needed them working the same day. He has since built and run software as a founder and lead developer, worked as a Salesforce consultant, and grown a YouTube channel on tech careers past one million views.

This book was last updated on September 4, 2026.

Connect on LinkedIn →

For teams

Put Fly.io on every desk at your company.

One licence, every book in the series, one link for everyone. From $250 a year.

See the team licences

Sources

Every source cited in Fly.io Made Crystal Clear. Each number matches the [n] marker next to that link in the book's "Go deeper" lists, so you can trace any claim back to its origin.

  1. [1]Fly.io documentation home
  2. [2]Fly Machines: comprehensive overview
  3. [3]Machines API reference
  4. [4]Fly.io raises $25 million Series D and appoints Scott Johnston as CEO
  5. [5]fly deploy reference
  6. [6]Fly.io networking: services
  7. [7]flyctl install
  8. [8]fly launch
Show all 39 sources
  1. [9]Health checks reference
  2. [10]fly logs
  3. [11]fly ssh console
  4. [12]Resource pricing reference
  5. [13]Fly.io pricing overview
  6. [14]Free trial documentation
  7. [15]Discontinued plans
  8. [16]Billing documentation
  9. [17]Cost management guidance
  10. [18]"Reliability: It's Not Great," Fly.io community forum, March 2023
  11. [19]"Fly.io Postgres cluster down for 3 days, no word from them about it," Hacker News, July 2023
  12. [20]"I was an early adopter of Fly.io. It is not production-ready," Hacker News, February 2024
  13. [21]"Reliability / Stability Pulse Check," Fly.io community forum, August 2025
  14. [22]"Fly Away from Fly.io," Jonas Scholz, June 2025
  15. [23]"Fly.io Reliability, SLA and Production Risks in 2026," Kuberns, 2026
  16. [24]Fly.io status page
  17. [25]fly.toml configuration reference
  18. [26]Fly.io secrets and environment variables docs
  19. [27]Fly.io multi-process apps and process groups docs
  20. [28]Fly App Scaling
  21. [29]OpenStatus: monitoring latency across Cloudflare Workers, Fly, Koyeb, Railway, and Render
  22. [30]Fly Volumes: overview
  23. [31]Fly.io regions reference
  24. [32]Fly.io: Private Networking (6PN)
  25. [33]Fly.io: Flycast
  26. [34]Fly.io: LiteFS
  27. [35]Fly.io: Managed Postgres (MPG)
  28. [36]Fly.io: Fly Postgres (unmanaged)
  29. [37]Fly.io: Upstash for Redis
  30. [38]Fly.io: Builders reference
  31. [39]Fly.io: Metrics reference

New books, in your inbox

One short note when a new Made Crystal Clear guide ships.
No spam, unsubscribe anytime.