Advanced · Spread work across servers and route around dead ones.
Load balancing
One server has limits. A load balancer sits in front of many and decides who gets each request, and with health checks it stops sending traffic to a server that's down.
Try this
Run each experiment in the simulator above, then tick it off.
Key ideas
Round robin
Take turns. Simple and fair when every request costs the same.
Least connections
Send to the least busy server. Better when request sizes vary.
Health checks
Ping every server; stop routing to one that fails and add it back when it recovers.
In your Pabbly workflows
You won't run a load balancer, but the pattern shows up: a fallback provider when your email or SMS service is down, or splitting one giant workflow's traffic across smaller ones so one slow path doesn't block everything.
Check yourself
0/2Get every question right to mark this lab done. Wrong answers let you try again.
Q1. Requests vary wildly in cost (10ms to 5s). Which strategy usually copes better?
Round robin ignores how busy a server already is. Least connections sends new work where there's room.
Q2. What do health checks do?
Without them, the balancer keeps sending a share of requests into a dead server.
Lab complete
Marked done on this device. On to the next one.
Watch next
Searches for “load balancer” on the channels worth your time.