Monitoring your bandwidth and staying inside your plan
Measure traffic, avoid overage surprises, and find what is actually consuming it.
Bandwidth is the easiest resource to overspend on by accident, because it scales with popularity rather than with your server.
What counts
Usually outbound traffic counts; inbound is often free or cheap. Serving a page, a video segment, or a download all count. Backups copied off-site count too — and they can be large and nightly.
Set an alert
Configure a bandwidth alert at, say, 70% of your allowance. Being warned is free; an overage invoice is not.
Find what is using it
# live throughput
iftop # if installed, per-connection
nload # interface totals
vnstat -d # daily totals per interface
# what is being served (nginx example)
awk '{print $1}' /var/log/nginx/access.log | sort | uniq -c | sort -rn | headAlso check the panel's traffic graphs for the trend — is it growing steadily (expected) or spiking (something changed)?
Common hidden consumers
Video/audio streaming — bitrate × viewers. The dominant cost on media hosts. Do the maths: Radio and video.
Backups — every night, off-site. Legitimate but budget for it.
Bot traffic — scrapers and attackers can consume real bandwidth. Rate-limit, block bad bots, and serve a CDN.
Large downloads — an un-cached 500 MB file downloaded often adds up fast.
Reduce it without hurting service
Cache — a page cache and a CDN cut origin traffic sharply.
Compress — gzip/Brotli for text, modern formats for images.
Serve static assets from a CDN — see Static assets and CDN delivery.
Right-size media bitrates — lower bitrate reaches more people for less.
Block abusive traffic — see Harden your VPS.
Planning for growth
If your trend says you will double within a month, move to a plan with more transfer before you hit the wall. Running out of bandwidth usually means a service that stops working, which is exactly the outcome the alert exists to prevent.
Related reading