A few months ago, I built this website following the concept of full-stack serverless, hoping to run WordPress on full-stack serverless instead of shared hosting. Later, I realized I was wrong.
As previously explained, full-stack serverless charges based on requests, making it the most economical way to run WordPress in low-traffic scenarios. When you have fewer than 100 users visiting your site daily, you only pay for 100 requests. Given the vast number of low-activity websites worldwide, this billing method allows these sites to exist on the internet at a low cost without worrying about server waste causing them to shut down. If no one visits your site, your server automatically shuts down.
This sounds reasonable, right? Yes, request-billed functions, request-billed caching, and request-billed databases. But I overlooked one crucial point: does one user visit equate to one database request? This is a very important point.
In reality, because a normal WordPress site uses many plugins from different vendors and developers, due to WordPress’s design and the varying performance of these plugins, a single page refresh can result in hundreds of database requests. And database access is more expensive than function access! This is precisely why we abandoned full-stack serverless WordPress: the cost was all spent on the database, even though the site itself didn’t have much user traffic.
Anyway, it was an interesting attempt and a lesson learned. I will keep the site open, hoping that when someone has the same idea, they can see this blog post.
I will continue to explore new things. This is my new website, an AI tool designed to help people with social media marketing: ReplyAIBot