The Scalability Paradox
It doesn’t matter as much as you think.
Engineers make their living when demand increases and supply needs to go up. We’re told stories when people built with and without scalability and how it made the difference between success and loss. We notice the times where popularity or disaster caused a spike. Pokemon Go. Zoom. GasBuddy. We admire the engineering prowess needed to build the most scalable platforms. Split.io’s clever SSE and wrapping SDKs. CDN providers. AWS.
We don’t hear the stories of when small was fine enough.
Famously, many of the websites on the internet are WordPress. Yes, a simple static site would be more scalable, cost less, and just as effective at showing “Home” and “About”. But heavy, ugly, MySQL+PHP WordPress wins. Not because of quality or magic open source pixie dust or ease of administration. It’s some combination of a wide and messy plugin ecosystem, low-skill self-service, and the somewhat cursed power of hacks that work well enough.
Scalability is a requirement for many engineers, not most sites. The proverbial wave of traffic usually doesn’t come. If it does, it’s much later, more predictable, or less costly to miss than you think.
The math usually doesn’t work out at small scale. Get it working and then get working on something else. When the wind shifts, then worry about scale.
Sincerely,
An Engineer
