Egress Costs, Explained

Networking in the cloud feels harmless until data starts leaving it. That’s when the meter shows and starts running. Egress is everything that goes out of your cloud - downloads, cross-region sync, CDN misses - and it’s the line item that sneaks up on teams the most.

The Basics

• Ingress - data coming into your cloud. Usually free.
• Egress - data leaving your cloud. Usually not free.
• East-west traffic - within a region or VPC. Often cheaper, sometimes free.
• Cross-AZ/region - traffic that hops zones or regions. Priced separately, and it adds up.

Why It Exists

Providers pay real money for backbone links, peering, and edge delivery. Egress pricing is how they recover those costs and manage network load. The more your traffic leaves their network or crosses geographies, the more you feel it.

The Usual Suspects

• User downloads - serving large files or media directly from object storage.
• Cross-region sync - active-active architectures that replicate data.
• Chatty microservices - services calling each other across zones.
• CDN cache misses - traffic falls back to origin more than you expect.

Why It Matters

Egress is a tax on popularity. The more people use your app, the more data leaves.

• On AWS, downloading from S3 to the public internet is billed, and inter-AZ traffic also has a cost.
• On Azure, outbound bandwidth and cross-region data transfer are separate line items.
• On DigitalOcean, Droplets include outbound transfer, but overages kick in once you exceed the included bandwidth.
• On Oracle, inter-region data transfer and internet egress are priced differently, and replication can multiply it.

The pattern is the same everywhere; moving bytes out = 💵 💲💸!

When Egress Is Worth It

Not all egress is bad. Sometimes paying the bill makes sense:

• Serving video or content globally without latency.
• Syncing regions for true disaster recovery.
• Multi-cloud strategies where resilience matters more than cost.

The key is knowing when you’re spending to create value vs. when you’re just leaking money.

The TAM Lens

Egress comes up in almost every review. It’s rarely about turning things off - it’s about moving the right traffic to the right place. Cache what you can, keep chatty services close, and be intentional about replication. Most surprises disappear when teams treat bandwidth like a design input, not an afterthought.

How to Stay Sane

• Put a CDN in front of static assets - push traffic to the edge and reduce origin fetches.
• Keep services co-located - same AZ or region for chatty paths.
• Be selective with replication - replicate what you need, not everything.
• Compress and chunk - smaller payloads, fewer roundtrips.
• Set budgets and alerts - watch egress like you watch CPU and storage.

Final Thoughts

Egress isn’t the villain - it’s the cost of delivering value to users. Plan for it early, design to minimize it, and you’ll keep performance high without the “why is our bill doing that?” moment.