A short guide to load balancing

The version of this that works is simpler than the version most people imagine. Everything we would tell a client about load balancing in the time it takes to drink a coffee.

Cloud work rewards teams who automate early and punishes teams who click through consoles. It is worth deciding this deliberately rather than inheriting whatever the last person set up.

What is actually at stake

Health checks decide whether load balancing helps or hurts. It is worth being explicit about, because assumptions differ quietly. The version that survives contact with a real deadline is the simple one.

What good looks like

Sticky sessions quietly limit how far you can scale. Small and consistent beats large and occasional here. The version that survives contact with a real deadline is the simple one.

Warning signs

Test what happens when one instance is slow rather than dead. This is the sort of thing that compounds, quietly, in both directions. Doing this properly once is usually cheaper than doing it approximately three times.

In practice

Operability is a feature, and it has to be built rather than bought. Three things worth confirming about load balancing before you move on:

• Someone can say what the current setup is without going to look
• Test what happens when one instance is slow rather than dead — and you know whether that is true here
• There is a way to tell whether the last change to this helped

Worth checking on your own setup before it becomes someone else's problem to fix.