🤖 Judoscale News: ⏳ Fly.io CPU Sharing Explained 🤯 (April 2025)
Greetings, Judoscalers! And welcome to the April edition of Judoscale News — full of April showers bringing May flowers and spring sunshine finally hitting Ohio ☀️. Love to see it 😎
It’s been a busy month for the Judoscale team, particularly in the traveling department! Between a couple of different conferences, some spring break trips, and even a retreat, the Judoscale team has been giving out t-shirts, making friends, and getting some R&R 😊
Here’s what’s in store this month:
- 🤯 CPU Sharing on Fly.io Explained!
- 🐍 Choosing the Right Python Task Queue
- 🔢 ICYMI: The Universal PaaS Pricing Calculator
- 🔐 Heroku SSL Revisited (2025 Edition)
- 🤤 We Have All the Platforms
Let’s do it!
🤯 CPU Sharing on Fly.io Explained!
Whenever we expand our autoscaling feature-set to a new platform we always like to run our own production application(s) there first. Probably no surprise — you should dog-food your own apps! But our results and findings on Fly.io were... interesting.
We found that the shared hardware tiers were significantly cheaper than other PaaS’s, but that, at the same time, they provided less horsepower than other platforms. So we needed to run quite a few more instances to compensate (eating away at those savings!). We finally figured out why. Read Adam’s Bluesky post here: (then ‘like’ and ‘subscribe’ 😆)
Jon’s take: it’s rad that Fly is clear, public, and succinct about what their specific CPU sharing algorithms / systems are and how you might expect those protocols to impact your application. I hope we see more of that in the coming years!
🐍 Choosing the Right Python Task Queue
We’ve probably all wrestled with background job systems at one point or another. It’s hard to build a compelling and useful application without some kind of background processing! But there are so many options. Tons of variables, tons of choices, and tons of time required to figure out which is the ‘best’! No thanks!
Instead, Jeff wrote a fantastic article that breaks down the landscape for you and talks through the tradeoffs simply. Celery, RQ, Huey, and even Dramatiq — whether you’re scaling a Django app or fine-tuning your FastAPI setup, this guide is packed with hands-on advice and hard-earned (🥴) opinions on performance, reliability, and production hurdles. If you’re building anything in Python, start here 💪:
🔢 ICYMI: The Universal PaaS Pricing Calculator
So maybe you’ve been a little platform-curious. Maybe you’ve heard the hype about Render and Fly.io over the last couple of years and fee like: 👀. Maybe you’re getting too big for your current platform and thinking about something like Amazon ECS, but actually getting a sense of how much something will cost on AWS before you actually use it is darn-near impossible. EITHER WAY! Adam built a fully featured PaaS Pricing Calculator to help you out!
Its goal is simple: compare the face-value, out-the-door costs for each platform in parallel once you describe your app setup.
So... thinking about jumping to Render or Fly? Maybe you’ll feel stronger about that idea once you see you’ll save fifty percent of your hosting costs! Or maybe you’ll feel less inclined if you see that you’ll double your hosting costs! The goal here is visibility. Check it out here!
🔐 Heroku SSL Revisited (2025 Edition)
Story time! Back in November I wrote “Cloudflare + Heroku SSL / Certificates Explained”, a quick but absolute source-of-truth guide to setting up SSL between Heroku and Cloudflare the right way. Well... there’s a second ‘right way’ now! And it’s quite a bit easier. Hence this new post from Adam, exploring how Team Orange and Team Purple now work together with some under-the-hood magic to make SSL certs just work. It’s interesting stuff! Give it a read here:
🤤 We Have All the Platforms
We’ll finish this newsletter off with a little bit of general excitement! The team here at Judoscale has been working for a couple of years on expanding our platform support. Not all that long ago we were simply Heroku-only. Now we support fully-featured, simple-setup autoscaling on five platforms:
Heroku, Amazon ECS, Render, Fly, and Railway. All Judoscalable! 🎉
👀 Sneak Peak: beyond the named platforms above, we’re also working on a little something-something... We want to make Judoscale available on any platform, and we think we’ve designed a custom setup that allows exactly that: Judoscale Everywhere. More to come in the future, but soon we’ll be fully platform agnostic 🤫🤯
If that sounds interesting to you, hit us with a reply to this newsletter. We'll take on a few beta testers soon!
It’s been a lot of fun to see everyone’s faces at conferences and meetups lately! We’re looking forward to more this summer. And, of course, we hope you all enjoy the T-shirts! 😁
— Jon & The Judoscale Team