Understanding AWS data transfer pricing is important for controlling network-related costs. Charges vary depending on whether traffic stays within a region, crosses to another region, or leaves AWS for the public internet.

There are three main categories of AWS data transfers we should be aware of: intra-region transfers such as cross-AZ communication inside a VPC, inter-region transfers between different AWS Regions, including cross-region replication, and internet egress when traffic leaves AWS to reach end users or external networks.

We shall consider each of these in turn while also highlighting common service scenarios where these costs appear, and point to the AWS pricing pages where each category is documented.

Intra-region data transfer refers to traffic that remains within a single AWS region but crosses Availability Zones. Transfers that stay inside one Availability Zone and use private IP addresses are free, but as soon as data moves between AZs, it's billed.

The standard charge is one cent per GB in each direction in most commercial regions, as documented on the Amazon VPC pricing page. This charge is easy to overlook because many AWS services rely on cross-AZ communication for high availability.

For this reason, if an application server in us-east-1a connects to a database in us-east-1b, every query and response generates billable traffic.

In addition, an application load balancer that routes requests across multiple AZs incurs cross-AZ transfer costs, and services like Amazon EFS or ElastiCache replicate data between AZs in ways that also add up over time.

Cross-AZ transfer is not necessarily bad. It represents a balance between achieving resilience and managing cost, and therefore distributing workloads across multiple AZs improves availability, but it also increases the amount of billed traffic.

For critical workloads, running across multiple AZs is a best practice for resilience, even though it increases costs. For other workloads, costs can be lowered by keeping resources that exchange a lot of data, like application servers and databases, in the same Availability Zone.

Having considered intra-region transfers, we shall now turn to inter-region transfers, which involve data moving between different AWS regions. Inter-region data transfer refers to traffic that moves between different AWS regions.

This traffic is billed separately in each direction. So, if data flows from us-east-1 to eu-west-1 and then back again, charges are applied for both paths.

The cost of inter-region transfer is higher than cross-AZ traffic inside a single region. Rates typically range from 2 to 9 cents per GB, depending on the region pair.

Now, unlike cross-AZ pricing, which is listed on the Amazon VPC pricing page, inter-region transfer is documented on the EC2 on demand pricing page under data transfer.

Even when services such as S3 or DynamoDB replicate data across regions, the billing follows the same region-to-region rates published under EC2.

Many AWS services rely on inter-region traffic. Amazon S3 cross-region replication is an example which creates copies of data in a second region for backup or compliance.

DynamoDB global tables synchronize data between regions to support multi-region applications, and Aurora Global Database replicates rights across regions to improve disaster recovery and read performance.

Even direct application traffic between EC2 instances in different regions is billed as inter-region data transfer. Inter-region transfer is often necessary for business reasons such as disaster recovery, latency optimization, or meeting regulatory requirements.

At the same time, because the rates are higher than intra-region transfer and billed per GB, costs can grow quickly for workloads that move large volumes of data across regions if you're not careful.

Having looked at both intra-region and inter-region transfers, we shall now turn to internet data transfer, also known as egress: the most expensive of the data transfer classes.

Essentially, internet egress refers to data transferred out of AWS to the public internet. This applies to any service that delivers content directly to end users or external systems, including Amazon EC2, Amazon S3, and streaming workloads.

Unlike intra-region or inter-region transfers, internet egress is billed at the highest rates. In most commercial regions, the first GB per month is free. After that, data transfer out is charged on a tiered basis.

To illustrate this tiered pricing model, let's look at us-east-1 region. The first ten terabytes each month are billed at nine cents per GB. The rate decreases slightly as the transfer volume grows, with the next 40 terabytes billed at 0.085 dollars per GB, and the next 100 terabytes at seven cents per GB.

These rates are documented on the EC2 on demand pricing page under data transfer, and the S3 pricing page provides the same model for objects delivered directly from S3.

I'm sharing these specific numbers to help you understand the real cost implications you'll face. However, keep in mind that AWS does update their pricing from time to time.

In practice, many workloads generate internet egress costs. Serve files directly from S3 buckets? You're paying egress for every download. Run web applications on EC2 that send responses back to users? More egress charges.

Where it really adds up though is with media and streaming applications where every single GB delivered to viewers hits your bill at full egress rates.

Having now considered intra-region, inter-region, and internet egress, we can move on to examining common patterns where these costs appear and how to identify them in your AWS billing data.

We have now considered the three main categories of AWS data transfer charges: intra-region transfers, where traffic crosses availability zones inside a region, inter-region transfers where traffic moves between different AWS regions, and internet egress where data leaves AWS for the public internet.

Understanding these categories is the foundation for controlling networking costs. While most charges fall into these three areas, there are also other factors that can influence billing, such as the use of public versus private IP addresses, NAT Gateways, and VPC endpoints.

We will explore those in later lessons when we turn to reducing unnecessary transfers and optimizing network design.

AWS data transfer pricing overview

About