Why Your Website Goes Down at the Worst Possible Time

The pattern repeats: a campaign goes live, an SMS blast goes out, results are published — and the site that was perfectly fine yesterday stops responding. Then it recovers on its own an hour later and everyone forgets until next time.

Usually it's the database, not the server

One slow query that's fine with ten visitors becomes fatal with a thousand, because each request holds a connection while it waits. The server looks idle, the site is dead. Adding RAM doesn't help; adding an index usually does.

The usual causes, in order

• Queries with no index, loading full tables per request
• No caching, so every visitor recomputes the same page
• Images served at full size to every mobile visitor
• Disk full — usually logs nobody rotates
• SSL certificate expired, which looks exactly like an outage

Know before your customers do

Uptime monitoring costs almost nothing and pings you when the site stops responding. Most businesses we start working with have no monitoring at all, and learn about outages from an angry phone call.

Test before the campaign, not after

If you're about to send an SMS to fifty thousand people, run a load test first. An hour of testing beforehand is cheaper than a failed campaign you already paid for.