EGRESS AND CLOUD COST REDUCTION
Why are you paying to move the same data twice?
Moving data into the cloud is free. Moving it back out, to users, other regions, or partner networks, is billed every time. It doesn't matter that it's the same response you sent a minute ago: every crossing gets billed.
Varnish Enterprise is a software-defined caching and origin shielding platform that stops those repeat charges. It sits inside your network perimeter so workloads never pull identical data across a billed boundary twice.
Why does cloud egress cost so much?
Your API keeps answering the same question
A response served to one user in another region costs one egress charge. Serve that same response to a thousand users in that region, and you pay a thousand times for data that never changed. Multi-region architecture that looks efficient in code is expensive on the network.
→ The fix: a regional cache that answers repeat requests locally, so only the first request ever crosses the boundary.
Your cache is missing more than it should
A tracking parameter, a session cookie, or a timestamp in the URL is enough to make one cacheable asset look like a thousand different ones. Every one of those "misses" reaches the origin and gets billed, even though nothing about the underlying content has changed.
→ The fix: normalize the cache key before the lookup, so the variations collapse back into a single cached asset.
A restart resets the meter
When an edge node reboots or an in-memory cache clears, the next wave of requests all miss at once and go straight to the origin. Routine maintenance turns into a billing spike, and often a reliability incident at the same time.
→ The fix: persist the cache to local disk, so a restart doesn't mean starting from zero.
Does this sound familiar? If you're trying to forecast unpredictable cloud spend or control costs in a multi-region setup, this is the infrastructure challenge Varnish can fix.
How does Varnish reduce egress costs?
Say you're serving images, models, or API responses from an S3 bucket, GCS, or Blob storage. Every request that hits that bucket gets billed, even for an asset you served an hour ago.
Put Varnish in front of it as a shielding cache: the first request pulls from the bucket and pays egress once. Every request after that is served from Varnish's local cache, and the bucket never sees it again.
01
Regional origin shields |
Read-through cache intercepts requests before they leave the perimeter. |
02
Cache key normalization |
Strips non-functional parameters so variations collapse to one cached asset. |
03
Horizontal cache peering |
Nodes check neighbors before the origin, so one fetch serves the cluster. |
04
Persistent local storage |
Cache survives restarts, so maintenance doesn't trigger a fresh fetch wave. |
Why don't CDNs or static replication already solve this?
Challenge 1
Cloud-native caching tiers
Locked to a single availability zone; no help once traffic crosses regions.
Challenge 2
Public CDNs
Absorb external traffic but leave internal and inter-region transfers untouched.
Challenge 3
Static data replication
Cuts egress but duplicates storage and adds sync overhead.
Challenge 4
Basic proxies without peering
Ten nodes with the same miss make ten separately billed trips.
The numbers
Egress fees convert from a variable monthly shock to a flat, predictable hardware cost.
99.8%
99%+
50%
"We were able to reduce our Azure costs up to 15 times, and the investment pays for itself. We also reduce the burden on engineering resources in terms of people having to manage this on their own, which could be up to one week savings every two months."
CTO
Automotive technology company
Varnish Enterprise
Slash egress fees. Shield origins. Control data transfer costs.
Varnish Enterprise is a software-defined caching and origin shielding platform designed to eliminate repetitive cloud data transfer fees. Operating as a highly programmable buffer tier inside your network perimeter, it handles high-volume read traffic locally so your workloads never have to pull identical assets across costly hyperscaler boundaries twice.
Resources and media
-
Creating an S3 shield with Varnish Enterprise
The S3 shield works not only for Amazon S3, but other solutions that implement the S3 API.
-
The secret to reducing cloud storage fees while improving performance
With Varnish we can build an S3 shield: a caching layer that will preserve the benefits of cloud storage while also solving the problems of cost and data locality.
-
Case Study: Nowcom
Nowcom uses Varnish for both frontend and backend caching to achieve up to 15x savings on Azure bandwidth costs
Next steps
Talk to our team to learn more about Varnish Enterprise today.
