Data egress waste is network spend that buys you nothing: bytes that cross a billing boundary needlessly, or the same bytes that cross it again and again because nothing caches them. Unlike compute, egress rarely shows up as a single big line item; it accumulates across thousands of small transfers, which is why it hides so well and why reducing it is often pure margin once you can see it.
This article sits in our complete guide to cloud rightsizing and waste elimination, the cluster pillar it links up to. Reducing egress waste is a Cut move under our See, Cut, Lock, Run method, and it pairs closely with reducing inter-service and inter-region traffic, which goes deeper on architecture-level flows.
Most clouds charge differently for traffic that stays inside an availability zone, crosses zones in a region, crosses regions, or leaves to the public internet. Egress to the internet is usually the most expensive, cross-region next, cross-zone smaller but high-volume. Map each flow to its boundary before you cut anything.
Step 1: Make the egress visible before you touch it
You cannot reduce what you cannot see. Start by attributing egress to its source. Cost and usage reports on every provider break network charges out by type, and flow logs (VPC Flow Logs on AWS, NSG flow logs on Azure, VPC Flow Logs on GCP) show which workloads talk to which endpoints and how much they move. The first pass almost always surfaces a handful of flows responsible for most of the spend: a backup copying cross-region, a logging pipeline shipping everything to a far endpoint, or a service in one zone calling a database in another for every request.
Step 2: Collapse needless cross-zone and cross-region chatter
The cheapest byte is the one that never crosses a boundary. Co-locate services that talk constantly into the same availability zone so their traffic stays intra-zone, and keep an application and its primary database in the same region. Many estates pay cross-zone charges purely because deployment spread services across zones for resilience without considering that every call now crosses a billing line. Balance resilience against cost deliberately rather than by accident. The architecture patterns here are detailed in reducing inter-service and inter-region traffic.
Want the egress flows mapped for you?
Our cloud cost audit attributes every network charge to its source workload, ranks the wasteful flows by dollars, and hands you a prioritized plan to cut them. On the performance model, you pay only from realized savings. No savings, no fee.
Book a cloud cost audit →Step 3: Cache and offload the traffic that must leave
For data that genuinely serves the public internet, put a CDN in front so repeat requests are served from the edge instead of paying origin egress every time. Egress from a CDN edge is typically priced below raw origin egress, and the cache hit ratio multiplies the saving. Use a private backbone or a peering and Direct Connect style link for predictable high-volume flows to your own data center, which is usually cheaper than metered internet egress at scale. The full playbook for this is in using a CDN and caching to cut egress bills.
Step 4: Compress, batch, and stop shipping what you do not need
Once the big structural flows are handled, shrink what remains. Compress payloads before they cross a boundary, batch chatty small requests into fewer larger ones, and filter logs and telemetry at the source so you are not paying to ship debug-level data you will never read. Observability pipelines are a frequent offender, covered in reducing logging and telemetry storage costs. Sampling and source-side filtering often cut that egress by more than half with no loss of signal.
| Waste pattern | Where it hides | Fix |
|---|---|---|
| Cross-zone request chatter | Services split across zones | Co-locate hot paths in one zone |
| Repeated internet downloads | Static assets served from origin | Put a CDN in front, cache aggressively |
| Cross-region replication | Backups and DR copies | Right-size frequency, compress, scope it |
| Verbose telemetry shipping | Logging and metrics pipelines | Filter and sample at the source |
Pricing relationships above (internet egress highest, cross-region next, intra-zone lowest) reflect provider models as of May 2026. Verify current per-GB rates and free-tier allowances in each provider's pricing pages before modeling savings, since egress pricing changes.
Watch the managed services that move data quietly
A large share of egress waste hides inside managed services that move data on your behalf without an obvious line item. A NAT gateway charges for every gigabyte it processes on top of the egress itself, so routing all outbound traffic through one when a gateway endpoint would keep object-storage traffic on the provider network is a common and expensive default. Managed databases with cross-zone replicas generate inter-zone traffic on every write. Load balancers spanning zones move request and response bytes across the billing boundary. Inventory these managed flows alongside your own services, because the architecture decision that created them was usually made for resilience or convenience with no view of the network cost it added. Often the fix is a routing change, such as adding a private endpoint for storage access, rather than touching the application at all.
The Cloud Storage and Egress Cost Playbook includes the flow-log queries and the boundary-mapping worksheet we use to attribute and rank egress spend. It is the downloadable companion to this method.
Make egress someone's number
Egress waste creeps back because no one owns it. Compute waste has a natural owner in the team that runs the service, but network charges are diffuse and easy to treat as a fixed cost of doing business. Break that by attributing egress to teams through tagging, so each group sees the bytes its services move and the dollars attached, and review the largest flows on the same cadence as the rest of the bill. When an engineer can see that a single chatty integration is costing real money every month, the fix usually follows without a mandate. This is the governance, or Lock, step of our See, Cut, Lock, Run method: once you have cut the obvious waste, keep it cut by making the cost visible to the people who can change it. The broader tagging and allocation discipline that makes this possible is part of any mature cost program.
The short version
Make egress visible by attributing it to source flows, collapse needless cross-zone and cross-region chatter, cache and offload the traffic that must leave, then compress and filter the rest. For the why behind the charges, read data egress charges explained. When you want the whole estate audited and cut at once, that is what our rightsizing and waste elimination service delivers.