Egress Traffic

Large internet services don’t run on the user’s device, but in huge halls full of computers called data centers. When you play a video there or download a file, data leaves that hall and travels to you. This outgoing data traffic is exactly what’s called egress traffic. The English word “egress” means exit. The counterpart is called ingress: data that flows from outside into the data center, for example when you upload a photo. The difference sounds technical, but is above all a question of money.

Why data going out is more expensive than data coming in

Companies that rent out computing power and storage space are called cloud providers. The largest are Amazon Web Services, Microsoft Azure, and Google Cloud. For almost all of them, incoming traffic is free, but outgoing traffic is not. You pay per amount of data transferred, usually a few cents per gigabyte. One gigabyte is roughly equivalent to half a feature film in medium quality.

For individual files, this is negligible. For a video service with millions of views, it adds up to millions of dollars a year. For streaming providers, egress is therefore a major line item in the balance sheet, right alongside licensing costs for films. AI companies are affected too: anyone shuttling massive training datasets back and forth between providers pays every time the data leaves the cloud.

Critics see this as a brake on competition. Shoveling data in costs nothing, but getting it back out does. Anyone wanting to move their data to a cheaper competitor gets a hefty final bill. Experts call this effect lock-in — you’re stuck with the provider. The EU responded with the Data Act and severely restricted switching fees; several providers subsequently dropped egress charges for migrations.

How egress is counted and billed

Billing is based on volume, not time. The provider counts the bytes that cross its network boundary outward and adds them up over the month. The price also depends on where the data is going. Within the same location it’s often free. Between two regions of the same provider it’s cheaper. Out to the open internet it’s most expensive.

Prices are also tiered: the first hundred gigabytes are often free, after which the price per gigabyte decreases as volume increases. This is precisely why egress bills are hard to predict. A bug in the code that repeatedly fetches data in a loop can generate four-figure charges overnight.

Technically, egress is reduced mainly by not sending data twice. This is done using a Content Delivery Network, or CDN for short: a network of caches that keep frequently requested content available close to users. If a video is delivered from such a cache, it doesn’t have to leave the main data center again. Stronger compression also helps, because fewer bytes carry the same information.

Egress in cloud bills and provider marketing

In the news, you’ll mostly encounter this term in disputes over cloud pricing. Providers like Cloudflare or Backblaze aggressively advertise charging no egress fees at all, taking aim at the three market leaders. Conversely, egress fees regularly come up in antitrust proceedings by the EU Commission and the UK regulator.

This also becomes practically relevant for small projects. Anyone running their own website or app on a cloud provider will find egress as its own line item on the monthly bill. A common mistake is confusing egress with connection bandwidth. Bandwidth describes how fast data can flow. Egress describes how much has flowed in total — it’s the amount that’s billed, not the speed.