Cloud Hosting Bandwidth Overage Fees, Explained
What bandwidth overage fees are in cloud hosting, the three ways providers handle traffic past the allowance, how to estimate yours, and how to cap the bill.
What bandwidth overage fees are in cloud hosting, the three ways providers handle traffic past the allowance, how to estimate yours, and how to cap the bill.
What US-based cloud hosting means in practice, the four places data leaves the country without you noticing, and the questions that get it in writing.
Seven steps to host a website on a VPS: pick a server, connect over SSH, secure it, install a web server, upload your site, point your domain, and add HTTPS.
How to upload a PDF, a photo, or a long paste to a free AI chat without an account, what happens to the file, and what the free tier won't do with it.
A bandwidth overage fee is what a cloud host charges when the data leaving your server in a month goes past the amount your plan includes. Data leaving is called egress, and it's the one part of a hosting bill that's set by your visitors rather than by you. A page that gets shared somewhere popular can double your month's transfer in an afternoon, and whether that costs you nothing, a few dollars, or a surprise invoice depends on one sentence in the provider's pricing page.
That sentence is what this guide is about: what counts as egress, the three different things providers do when you pass the allowance, how to estimate your own transfer before you buy, and how to make sure the bill has a ceiling. If you're comparing hosts more broadly, the checklist in How to Evaluate a Web Host and Pick the Best One puts bandwidth in context with the rest.
Bandwidth is the speed of your server's network connection, measured in megabits per second. Egress is the amount of data that leaves the server over a month, measured in gigabytes. Overage fees are almost always about egress, the monthly amount, and almost never about speed. Inbound data, what your server downloads, is usually free.
The three numbers on a pricing page and what each one means for your bill:
A plan can have a fast port and a small allowance, which serves each visitor quickly and runs out of included data sooner. Or a slow port and a big allowance. For a small site, the allowance is the number that matters.
Past the allowance, a provider does one of three things: bills you per gigabyte at a published rate, drops your speed to a lower tier for the rest of the billing period at no charge, or stops serving traffic until the next period. Which one your plan does is the single most important bandwidth fact to know before you buy.
The three, with what each means in practice:
The six jobs a free AI chat does well for a small business, the two it shouldn't touch, and how to pick one without handing over a card or your customer data.
The right choice depends on which failure you'd rather have. A store would rather pay for the spike than slow down during it. A documentation site would rather slow down than pay. The provider's default should match yours, and if it doesn't, you want a switch.
Multiply your monthly page views by the average size of a page, add your downloads, and add a third for assets, bots, and the things you forgot. A text-heavy page with a few images is around a megabyte. A page view count you don't have yet, assume ten thousand and revise after the first month.
Worked through for a small site:
That fits inside most entry allowances with room. Where it stops fitting:
After the first month, read the transfer graph in your provider's dashboard and replace the estimate with the number.
Put a cap on it, serve big files from somewhere else, and make your pages smaller. The cap is the one that matters: either pick a provider whose default is to throttle rather than bill, or set a spending limit if the provider offers one, so a spike has a ceiling you chose.
Four things in order of how much they help:
The limitation of all four is that none of them helps with the traffic you want. If your site is popular because people are downloading your product, the egress is the business, and the only fix is a plan sized for it. On Carpathian, that's the larger network tiers or the unlimited data add-on, priced on the same page as the base plans. On a per-gigabyte provider, it's a budget line.
No. Per-gigabyte billing is the norm on the large public clouds and on many smaller ones, but a number of providers include a fixed allowance and throttle past it, and some bundle a network plan across every server you have rather than per instance. Read the policy on the pricing page, not the allowance in the headline.
The question to ask, in one sentence: "If I use twice my allowance next month, what is the bill and is my site still up?" A provider that can answer that in one sentence has a policy. One that points you at a calculator has a rate.
Open the pricing page of whichever host you're considering and find the sentence that says what happens past the allowance. If it's a per-gigabyte rate, decide whether you can absorb a spike. If it's a throttle, decide whether you can absorb a slow day. Carpathian's version of that sentence is on https://carpathian.ai/pricing/cloud, and a server with the throttle default is a few minutes away at https://carpathian.ai/signup?service=cloud.