June 19, 2023
Websites used to go down with more regularity than they do now. Twitter famously had the fail whale. But I mean more like “normal” websites, like this one. If Daring Fireball linked to a site and it went down, that community called it getting “Fireballed” and there was exotic solutions. If Slashdot traffic took down a site it was “slashdotted”. There was the “Digg Effect” and “Reddit Effect” and the like.
You don’t really see that anymore. Why?
I asked recently, and here’s some ideas.
- CDNs. Even if just the media assets for a website are CDN’d, which is certainly a best practice that lots of sites take advantage of, that takes a lot of the load off the main web server. More elaborate CDNs like Cloudflare help keep sites up even moreso, as they protect against traffic as intense as a DDoS.
- Hosted Services. Many websites these days aren’t independently hosted but hosted by a big site-hosting service like Squarespace, Wix, WordPress.com, or the like which can absorb the traffic. There are fewer independently-hosted sites.
- Even with independently-hosted sites, the servers and server software is better now. Those hosts may be on beefier and more scaleable infrastructure themselves.
- SRE is a job now.
- It’s more embarrassing to go down, and the market doesn’t tolerate it. The web is too central to business.
- More serverless. As in servers that are entirely handled by an infrastructure company and often involve serving static assets and enforceably-small-and-fast server-side nuggets.
I think all of those things are true and each of them contributes to some degree toward the general feeling that sites don’t go down much to traffic spikes anymore.
Related
🤘
ncG1vNJzZmibmKe2tK%2FOsqCeql6jsrV7kWlpbGdga3xyhY6wn7JllKS7tXnSoqueq12cvG6wzrClZqWlmLVurc2ypKiqlWQ%3D