Judoscale: One-Click Autoscaling for any hosting platform ✨

Howdy, Judoscalers! Spring is springing, temperatures are warming up all over the place, and Judoscale is bringing the news right to your inbox! Here’s what we’re covering this time around:

• 🔄 We’re Going on Tour... Starting With Heroku
• 📈 Busy Jobs Display
• 📝 A Note on AI Writing
• 🖥️ Long Live Staging?
• ➡️ Heroku Add-On to Direct-Account Migration Tool
• 📝 Latency-based Celery Queues in Python

Lots of good stuff… let’s talk about it!

🔄 We’re Going on Tour... Starting With Heroku

As we mentioned at the end of our last post, “Heroku: What’s Next”, we’re taking Judoscale ‘on tour’:

…we’re going to migrate the Judoscale production application, including all traffic, DNS, configs, background workers, etc, to each of Heroku’s competitors, one at a time, and document every step along the way for you all.

And we felt the best way to start our tour was actually with a reflective and introspective look at Heroku itself! Not in the sense that we want to set a bar for other platforms to reach, but instead that we wanted to develop a rubric we could use across all platforms. And we did! We call it the Friction Model. This first article explains that model and applies it to Heroku as both an experiential reflection and objective analysis. Check it out here:

📈 Busy Jobs Display

We shipped a neat little visual feature in the Judoscale dashboard app this month, particularly for users who utilize our long-running jobs tooling. As a reminder, that feature-set basically prevents Judoscale from downscaling your worker processes if they’re actively processing a job, which can be helpful for apps with jobs that run for a long time! For those teams, we now show that “busy” state directly in the Judoscale UI as a helpful awareness metric. So, when you have this checked:

You’ll now see a new line in our metrics charts showing you when, exactly, you had longer jobs running:

Nice! ✨

📝 A Note on AI Writing

You might’ve noticed it on our last blog post (“Heroku: What’s Next”), but after doing some thinking and reflection of my own (“The Right Answer is to Disclose”) I’ve opted to add a little ‘Note on AI use’ button next to my name, which opens up the banner and text shown below. It’s a strange world we live in where this feels necessary, but the goal is to inform our readers that we care about what we’re writing and that our opinions really are our own. Also, I love the em-dash, deliberately use it incorrectly, and have for years — AI will not take that from me!!

Anyway, what do you think? Did you notice this before? Does this mean anything to you as a reader? Do you appreciate this kind of disclosure or do you not care? Is yellow the wrong color? 😜 (Just reply to this email!)

🖥️ Long Live Staging?

One thing we were mulling over as we prepared for the Judoscale Tour was what we should expect and implement regarding the “staging server” vs. “production” workflow. Specifically, while our own use of staging vs. prod isn’t actually a great representation of how most apps work (Judoscale staging actually autoscales Judoscale production!), we were thinking about what most people do for staging vs. production — including the “main auto-deploys directly to prod” setup. Good thing we hadn’t done that yet! Check out Adam’s story here:

➡️ Heroku Add-On to Direct-Account Migration Tool

After reading our aforementioned 'On Tour: Heroku' article, you might be less worried about migrating to a new host. That’s great! But while you remain on Heroku, you might as well save a little cash for free in the meantime! We recently created a direct-billing option for our customers on Heroku which comes with a discount on your monthly subscription. Same service. Same Judoscale. Fewer dollars! After setting up the direct-billing variant of our Heroku integration, we’ve now completed a simple migrator that can move your add-on-style account to direct-billing with ease. Check it out here:

📝 Latency-based Celery Queues in Python

Don’t think we’re leaving Python out this month! Did you know that most Celery setups don’t fail because of scale, but because of how their queues are defined? Jeff helps walk us through the right way to avoid a tangled mess of queues, expectations, and the dreaded “why is that job not running, it needs to run AS SOON as we push it!!” situations. There are some real keys to a successful Celery setup sprinkled through this article. Grab a fresh cup of coffee and give it a read here:

That’s the Judoscale News for this month, friends! Hope you all have a smoothly autoscaled month ahead and we’ll catch you on the next roundup. 💚

— Jon & The Judoscale Team