Why I Evangelize Static Sites As A Seasoned Engineer

Have you ever had to explain to a customer that the way they were using something was what was causing their problem? It's usually awkward, and really hard not to sound condescending or dismissive, but sometimes you just have to tell someone hey, you're not going to get that thing to do what you're trying to get it to do.

That's what it was like working in the hosting industry in the early 2000s when people wondered why their WordPress site was suspended on a shared server for over-using resources. And that's when I hyper-fixated on static sites as the only really naturally "scale-able" solution.

They still are.

It wasn't that these blogs were being bombarded with traffic, its that several of them getting crawled by Slurp on the same server would cause a cascading starvation of resources. When these sites were crawled regularly and received substantial traffic, they quickly needed load balanced clusters.

Unfortunately, you're just as, if not more likely to become a victim of your own success if you just use a database driven solution in 2025.

To Begin, I Actually Like WordPress - A Lot!

WordPress might be the most successful fork in the history of free software.

That's not a trivial thing to say about anything. WordPress became what it is because someone saw something in it and knew that they had the unique vision to make it a tool that people would love to use. And, for a very long time, if you didn't want to hand-edit a whole lot of HTML, it was basically your only reliable option to launch a content site on a budget.

But its design is fundamentally at odds with high-volume concurrent access, has always been that way, and will always be that way. This goes for almost every site that unnecessarily puts a dozen database queries between every visitor and the page they want to see.

My early career in tech involved many sleepless nights because of PHP, MySQL, WordPress and Apache. They ruined relationships, caused pets to run away and my hair to fall out faster. But, I digress.

WordPress Has Always Been Problematic To Host

Imagine that you have a fun evening with some friends planned. You're going to meet your friends "Mork" and "Mindy" for appetizers at a new Italian place.

You get showered, brush your teeth, style your hair, put on a nice outfit, some accessories and then you travel to the bistro and give your friend Mindy a great big hug. You see Mork over at the bar, so you go home, shower, brush your teeth, put on a nice outfit and then travel back to the bistro and give Mork a great big hug.

Wait, that's really wasteful, right? Why can't you just get ready once and meet more than one friend when prepared?

The same logic applies to database-driven sites that require a dozen queries for every connection that they serve, which is exactly the same for everyone else.

The Best Cache Is Linux + A File System

If you want to see Linus get very animated and you're afraid to talk about Rust in the kernel, just mention O_DIRECT. Go on, I dare you. Sneak up behind Linus and say BOO! O_DIRECT !!! Actually, don't do that; he knows martial arts and it's a bad idea to mess with people to prove silly points.

But back when all this insanity of using relational databases to store everything in the world got started, he (like many) was flabbergasted by what the RDBMS enthusiasts wanted out of the kernel: a way to bypass all buffers and operate "on the metal" when they opened and wrote to the files that actually backed the data in their databases.

The problem, as they saw it, was that I/O from buffers could not be guaranteed when systems were less than stable. And, they were right, but the elephant in the room is that it was their database servers, and the way they were being leveraged so indiscriminately, that was causing all of the server instability to begin with.

Web hosts built offerings around how many databases you could have. Bad engineering decisions is what drove growth back then, what can I say? But the fact that people are still starting out likely to topple over upon success is still extremely frustrating to me.

"Caching" only goes so far.

Some might argue that WordPress and others do have caching plugins. They're correct, but PHP needs to be invoked in order to serve a static file. Sure, caching helps, but if it's not just a plain text file that a lightweight web server can deliver, then it's overpowered.

Sure, some sites need to be connected straight to a database, like the sites where we pay our bills, do our banking, check our order history, mindlessly scroll, etc. When each page varies greatly depending on who and where you are, complexity makes sense.

To serve blog posts that seldom if ever change does not require complexity, it requires 12 lines of TS server code and a file system. Generating those pages is a whole other matter, and where efficient tools like Lume come in.

I'm not saying get rid of WordPress. I'm saying stop using it as a web server and stop continuously re-generating the same exact pages on every request.

It really is as silly as showering and changing before greeting each friend in a line.

Efficient Servers Turn The Tables On Bad Actors

There are many great hosts that are doing some really creative things with WordPress and many other open source content platforms, and giving back not just money, but contributions in time and effort to improve the software. Those companies know who they are, and how much we love them.

I'd like to talk about the other 97 percent, because a substantial amount of what's out there is snake oil.

There are a lot of not so great companies that prey on inexperienced site owners and the best way to fight them is to make people not need them in the first place. I think embracing the idea of the database being for creation and generation with comments and analytics on the side is sound, because so many pages out there don't need the overhead of being created on every connection.

But that adds friction, and when you've taken money, your bottom line can't be very altruistic. So I see it as something that's important, but that the community needs to deliver.

Because continuing to watch new publishers become victims of their own success seems kinda dumb. Can't we come up with something easy for people that also won't explode at the first sign of success?