You Might Be One Rug Pull Away From Crippling Infrastructure Bills

Somewhere around the early 2010s the hot tech news was companies moving their tech off self managed “On Prem” hardware and migrating to the cloud. It was a bright future with promises of high availability, low latency and easy to manage infrastructure. Job titles like Server Admin were on the decline, with the promise that developers could do the procurement and provisioning of the resources they need without those pesky, perpetually annoyed Server Admin Gremlins.

All was not rainbows and butterflies though, have you ever tried to custom build your own custom AWS architecture for a moderately complex application? I’ve spent many-a-night bashing the keyboard with my forehead trying to get a CodeBuild deployment to connect to an RDS database to make migrations run during CI. Does AWS make their error messages purposefully cryptic?

Turns out the job of the Server Admin was actually kind of complex, even when you’re only provisioning your resources through a PaaS like AWS. This gave rise to the DevOps specialization. Around 2015 DevOps Engineers were the rising stars. Building out complex Rube-Goldberg like architectures of loosely tied together PaaS services to make CI pipelines work like greased conveyer belts smoothly pushing code from developers to complex production environments.

Where there’s friction there’s opportunity. Dev Ops has been one of the most notorious areas of friction for development teams in the past 7 years or so. This has given rise to a veritable slew of developer focused PaaS solutions that promise to reduce the insane complexity of managing your own cloud infrastructure. Anything from auto-scaling database systems, to plug-and-play authentication solutions.

Each of these products promising amazing features. Netlify making it trivially easy to deploy code from scratch to production without fussing over EC2 ram requirements, Firebase making realtime multiplayer applications available to your average developer and many many more. Venture Capital funding pumping billions of dollars into these startups meant you could find a generous free tier, or at least a trivially cheap bottom tier unless you’re running a Netflix.

These PaaS have been busy enabling developers to once again bypass the new smelly gremlins in the DevOps teams, and spinning up their own services with a hugely reduced surface area of expertise for managing these services. These new options popping up nearly everyday, you could trick yourself as a developer into thinking the good times were really going to keep rolling on.

All good things come to an end though, and thus, the end of generous PaaS infrastructure companies is nigh. Venture Capital is busy pulling all of their funding from PaaS to chase down the new hotness; AI. This radical shift in money flow has seen many PaaS companies pulling the rug out from under their problematic free tiers. Like a drug dealer, they get you hooked with a little sample, then when you’re hooked, jack up the prices.

Recent examples include:

• MongoDB Atlas: MongoDB cut back its free tier, limiting it to smaller clusters and fewer regions. This frustrated developers who had used the platform for more robust and distributed applications​ (Smashing Magazine).
• Trello: Trello, the popular project management tool, reduced the number of free boards users could create, pushing them toward their premium plans. This created dissatisfaction among small teams and freelancers ​(DEV Community).
• Netlify: Netlify, a popular web hosting and automation platform for developers, altered its free plan by placing stricter limits on bandwidth and build minutes, prompting developers to either scale back their use or upgrade ​(CrafterCMS).
• CircleCI: In 2022, CircleCI drastically reduced the number of free build minutes, which led to frustration among smaller teams and open-source developers who relied on the service for continuous integration​(Built In).
• Zapier: Zapier reduced its free tier by limiting the number of Zaps (automated workflows) and restricting premium app integrations, pushing users towards paid tiers. This was particularly frustrating for developers who had set up multiple automation flows ​(Smashing Magazine)​ (CrafterCMS).
• Typeform: Typeform significantly cut back on its free tier, reducing the number of responses and surveys users could collect before being forced into a paid plan. This angered users who relied on the service for small-scale feedback collection ​(Built In).
• Firebase: Firebase, a platform by Google for building mobile and web apps, reduced its free tier quotas for services like Firestore and hosting, which affected developers working on small applications​ (Smashing Magazine).
• SendGrid: SendGrid, an email marketing service, scaled back its free plan, limiting the number of emails that could be sent each month. This forced users, especially small businesses and startups, to look for alternatives or pay for their services​ (ALMtoolbox).
• Algolia: Algolia, a search-as-a-service platform, reduced its free tier, limiting search queries and forcing developers with high usage to pay. This was especially frustrating for small startups that had relied on the free plan for scalability ​(Built In).
• Bitbucket: In 2021, Bitbucket announced changes to its free plan, limiting the number of private repositories and build minutes, pushing users toward their paid offerings. This left many open-source contributors frustrated ​(CrafterCMS).

This trend brings up some unsettling questions.

• Am I using any infrastructure that will be crippling if they “rug pull” my tier?
• Have I done my due diligence in making sure my applications aren’t overly reliant on services that don’t have a long history of price stability?
• Do I vet these services and assess the risk of becoming “hooked” on them causing an overly expensive or complex migration?
• Does it make sense to hire a DevOps engineer to manage these types of services on more primitive platforms like AWS where the cost control is more predictable?

Finally, what does the future hold? Will we see a resurgence of DevOps job postings as companies get more saavy and want more control over their future costs? Or will developers have to pick up more of the DevOps slack somehow?

What are your thoughts?