Most cloud waste is not dramatic.
It is small amounts, charged month after month, on things nobody is using any more. Because each line looks minor, it survives every budget review and quietly adds up. Here are the usual places it hides, and a fast way to check each one. None of this needs a special tool, an afternoon with your provider’s console finds most of it.
Machines left running idle. Development, test, and staging environments left on overnight and at weekends are the most common leak. Look for anything running at near-zero use, and automate the shutdown so it does not depend on someone remembering.
Forgotten instances. The server spun up for a launch two years ago, still charging you every month. List everything that is running and match it to an owner. Anything nobody claims is a candidate to switch off.
Oversized machines. Plenty of instances were provisioned large to be safe and now run at a fraction of what they can handle. Compare each machine’s real use against its size, and step it down where there is clear headroom.
Orphaned storage. Disks, volumes, and snapshots often outlive the thing they belonged to. Look for storage that is not attached to anything, and old snapshots nobody would miss.
Data moving around. Traffic leaving the cloud, or crossing between zones, adds up on the networking lines of the bill. Egress fees are falling, waived for departing customers in 2024 and banned across the EU from January 2027, but everyday transfer still costs, so it is worth knowing where yours goes.
Idle managed services. A load balancer, a gateway, or a database spun up for a test and never removed keeps charging by the hour. Check the per-hour services you are not actively using.
This is the ordinary drift of a setup that keeps changing while old things stay switched on. A regular look through these six catches most of it, and the money you stop spending is yours to keep.
