WordPress Caching: All You Need To Know

April 1st, 2021

Here’s Ashley Rich at Delicious Brains writing about all the layers of caching that are relevant to a WordPress site. I think we all know that caching is complicated, but jeez, it’s a journey to understand all the caches at work here. The point of cache being speed and reducing the burden on the worst bottlenecks and slowest/busiest parts of a web stack.

Here’s my own understanding:

  • Files can be cached by the browser. This is the fastest possible cache as no network request happens at all. Assets like images, CSS, and JavaScript are often cached this way because they don’t change terribly frequently, but you have to make sure you’re telling browsers that it’s OK to do this and have a mechanism in place to break that cache if you need to (e.g. by changing file names). You very rarely cache the HTML this way, as it changes the most, and file-name-cache-busting of HTML seems more tricky than it’s worth.
  • Files can be cached at the CDN level. This is great because even though network traffic is happening, CDN servers are very fast and likely geographically closer to users than your origin server. If users get files from here, they never even trouble your origin server. You’ll need a way to break this cache as well, which again is probably through changing file names. You might cache HTML at this level even without changing file names if you have a mechanism to clear that cache globally when content changes.
  • The origin server might cache built HTML pages. On a WordPress site, the pages are built with PHP which probably triggers MySQL queries. If the server can save the result of the things that have already executed, that means it can serve a “static” file as a response, which it can do much faster than having to run the PHP and MySQL. That’ll work for logged-out users, who all get the same response, but not for logged-in users who have dynamic content on the page (like the WordPress admin bar).
  • The database has its own special caching. After a MySQL query is executed, the results can be saved in an Object Cache, meaning the same request can come from that cache instead of having to run the query again. You get that automatically to some degree, but ideally, it gets wired up to a more persistent store, which you do not get automatically

Phew. It gets a little easier with Jamstack since your pages are prebuilt and CDN-hosted already, and in the case of Netlify, you don’t even have to worry about cache-busting.

But even as complex as this is, I don’t worry about it all that much. This WordPress site uses Flywheel for hosting which deals with the database and server-level caching, I have Cloudflare in front of it with special WordPress optimization for the CDN caching, and roll-my-own file-name cache busting (I wish this part was easier). I’d certainly trust SpinupWP to get it right too, given Ashley’s great writeup I’m linking to here.

The original post WordPress Caching: All You Need To Know.