Guide to Cloud Egress Costs for Media Archive Decisions

When your team is likely to restore an archive often, multiple outside teams will review it, or your team will use it as the source for new deliverables, storage price alone isn't a safe basis for choosing a cloud tier. Build the archive model around data movement first. The cheap line item is usually dollars per TB per month, but the expensive surprise is what happens when editorial, VFX, localization, legal, marketing, or a distributor needs the media back. For media teams, egress is the charge for moving data out of a cloud provider’s network or across a billing boundary. Downloading camera originals from object storage to your facility can trigger it. Sending masters to another cloud provider can trigger it. Moving files between regions can trigger it. In some architectures, even internal-looking transfers through gateways, peered networks, or managed services can create network charges.

Egress is the cost event that happens when media leaves a provider boundary.

That matters because teams usually make archive decisions when a project is quiet, but the costs show up when the project becomes urgent again.

The archive decision is really a restore decision

Most archive conversations start with, “Where can we store 200 TB cheaply and safely?” That's the wrong first question. The better question is, “What will we need to pull back, how often, how fast, and who needs it?” A wrapped feature, an unscripted season, a commercial campaign, and a sports archive don't behave the same way. They may all be “archive” on a spreadsheet, but their restore patterns are different. Common restore patterns include:

A full project restore for a recut, remaster, or conform
Selective pulls of camera originals for VFX, legal, clearance, or restoration
Review access to masters without downloading source media
Delivery of mezzanine files to distributors, platforms, agencies, or international partners
Disaster recovery after deletion, ransomware, or facility loss

Each pattern has a different cost shape. A full restore is easy to model because it's mostly TB out multiplied by the provider’s egress and retrieval charges. Selective pulls are harder because teams often retrieve more than they finally use. Review workflows can be either cheap or expensive depending on whether people stream proxies, download mezzanines, or repeatedly pull originals. Cloud archive is often the right offsite copy, especially when teams are distributed or when physical drive handling is becoming a mess. Stop pretending that the storage rate is the total cost.

What cloud providers charge for

Cloud storage pricing is layered, and the line item everyone notices is storage, which providers usually quote as dollars per GB per month. Egress is only one of the other layers, but it's often the one that changes the archive decision. When comparing storage options, include these cost categories:

Storage capacity, which providers charge by GB or TB per month
Retrieval fees, especially for colder archive tiers
Egress fees for moving data out of the provider or region
API and operation fees for listing, reading, writing, restoring, and deleting objects
Minimum storage duration, such as 30, 90, or 180 days
Early deletion charges if your team removes media before the minimum term
Restore priority charges for expedited access
Temporary restored copy charges for archive tiers that require a restore before download
Engineering or assistant editor time to find, stage, validate, and transfer files

Those last two are easy to miss. In cold archive tiers, “restore” may not mean “download.” It may mean “make a temporary accessible copy, then download it.” If nobody models the temporary copy window, the transfer tools, and the person babysitting the job, the archive looks cheaper than it's. For post teams, time is also a real cost. If a tier takes 12 hours or 48 hours before the first byte is available, that might be fine for disaster recovery. It isn't fine when a colorist, editor, or client is waiting on a shot.

A simple model for media archive TCO

Use a model that reflects how your media actually moves.

Archive cost grows from movement, delivery, restores, and handling—not storage alone.

Start with this yearly cost structure:

Annual archive cost =
  storage cost
+ planned restore cost
+ unplanned restore allowance
+ distribution and delivery cost
+ migration or exit allowance
+ operator and engineering time

For a first pass, model by project type rather than by the whole company. A scripted series might have one cost profile. A brand studio with constant campaign reuse might have another. A documentary archive with heavy search and clip licensing might have another. Here is a simplified example. Assume you archive a completed project with:

50 TB of camera originals, audio, project files, graphics, and masters
5 TB of mezzanine and review files your team is likely to reuse
500 GB of proxy or access copies
You expect one full restore every five years
Four selective restores per year averaging 1 TB each
External delivery of 2 TB per year

Now compare two broad storage approaches:

Cost area	Low-cost cold tier	Warmer or no-egress tier
Monthly storage	Usually lower	Usually higher
Retrieval fee	Often applies	Lower or none
Egress fee	Often applies	Lower, capped, or none depending on provider
Restore delay	Can be hours to days	Usually faster
Minimum duration	Often longer	Often shorter
Best fit	Rare disaster restore	Reuse, delivery, and remote access

The cold tier might win if the project almost never comes back. The warmer tier might win if producers, clearance teams, or localization vendors keep asking for files. The no-egress or egress-friendly provider might win if the archive doubles as a distribution source. This is why a $1/TB/month archive tier isn't automatically cheaper than a $6/TB/month tier. If the cheaper tier charges to retrieve, charges to egress, enforces a long minimum duration, and requires extra staging, it may still be the wrong place for media that your team touches.

Treat originals, working media, and access copies differently

Teams commonly make the mistake of putting every file from a wrapped project into the same cloud class. Camera originals, edit proxies, masters, project files, subtitles, audio stems, and marketing pulls don't have the same access pattern. A better archive design separates the archive into functional layers:

Separate media into layers so each file type sits in the storage class that matches its use.

Preservation copy: original camera files, production audio, final project files, final masters, color decisions, mix stems, and critical metadata
Working restore copy: mezzanines, plates, graphics, and other files likely to be reused in finishing or versioning
Access copy: proxies, screeners, thumbnails, transcripts, logs, and searchable metadata
Delivery copy: platform-specific masters or packages that may need to go to partners repeatedly

The preservation copy can live in colder storage if restore time is acceptable. The working restore copy often belongs in a warmer tier or on local nearline storage for a defined period after wrap. Keep access copies small and cheap to stream or download. Place delivery copies where repeated outbound transfer doesn't punish you.

Archive layer	Typical media	Best placement	Main egress concern
Preservation copy	Camera originals, production audio, final masters, project files, metadata	Cold cloud, deep archive, LTO, or another durable offsite copy	Large restore events are expensive if the whole project must come back quickly
Working restore copy	Mezzanines, plates, graphics, stems, commonly reused elements	Local nearline, warm cloud, or shared storage near editorial and finishing	Repeated downloads can cost more than keeping the working set closer to the team
Access copy	Proxies, screeners, thumbnails, transcripts, logs, searchable metadata	Low-cost online storage, MAM, review platform, or streaming-friendly tier	Poor access copies force users to restore originals just to inspect content
Delivery copy	Platform masters, packages, marketing assets, localization files	Egress-friendly cloud, CDN-like distribution layer, or partner delivery platform	Repeated partner downloads can turn outbound transfer into the dominant cost

This split is especially useful for editorial teams. Editors rarely need to pull 50 TB of originals just to answer a question. They need to search, view, compare, and identify the right file. If the archive has proxies, thumbnails, transcripts, and reliable metadata, the team can make decisions before touching expensive source media.

Keep active and recently wrapped work close to editorial

Hybrid storage is still the normal answer for many post teams because full cloud-native post isn't always practical. Fast NAS or SAN remains a strong fit for active editorial, especially where multiple editors are cutting HD, 4K, or 8K media on tight schedules. Cloud becomes more useful for offsite protection, remote collaborators, burst workflows, and archive access. Your team should define the boundary between “active” and “archive” as a policy. For example, a facility might use:

Fast local shared storage for current editorial and finishing
Local nearline storage for recently wrapped jobs and likely pickup work
Cloud object storage for offsite archive, remote access, and disaster recovery
Cold cloud or LTO for deep preservation copies that rarely need to move

This approach reduces egress because teams aren't repeatedly pulling the same working files out of the cloud. When a show is still in promo, localization, awards, or revisions, the likely working set belongs local or in a warmer shared tier. Move it deeper only when the restore pattern truly slows down. A useful test: when your own team downloads the same file from cloud archive more than once, your team probably should have staged it locally, cached it, or created a smaller access copy.

Egress is a workflow design problem

Teams often blame egress costs on the provider, but the workflow usually decides how painful they become. The most expensive pattern is repeated point-to-point downloading: one person restores a master, another vendor downloads it again, a producer downloads it again, then a distributor gets another copy from the same origin. The provider sees repeated outbound GB, and your invoice does too.

Restoring once to a shared cache can reduce repeated pulls from the archive origin.

Better patterns reduce repeated origin pulls:

Use proxies or mezzanine review files instead of originals for search and approval
Stage restored files once to a shared location instead of having every user download separately
Cache frequently requested files near the team that needs them
Keep cloud processing in the same region as the storage when possible
Use CDN-like distribution for files that go to many external viewers or partners

CDN-like distribution also works beyond public streaming. The same idea applies to private media delivery: keep the origin protected, push repeated access through a distribution layer, and avoid charging the archive bucket every time someone needs the same file. For screeners, marketing assets, training libraries, and frequently delivered masters, this can change the economics quickly. For large catalogs, delivery cost can dominate storage cost. If a platform, brand, or studio archive sends many TB per month to viewers, vendors, or partners, the storage provider’s egress model may matter more than its storage rate.

Comparing providers without getting fooled

Provider comparisons get messy because every vendor uses a slightly different pricing model. Some major cloud providers offer very low deep archive storage, but charge for retrieval and egress. Some S3-compatible providers charge more for storage but include, waive, cap, or reduce egress. Some pair with CDN networks in ways that make delivery cheaper. Others are attractive because storage is close to compute, or because their security controls or procurement terms fit the organization, even if raw egress isn't the cheapest. When comparing providers, avoid a single “price per TB” column. Build a comparison around the way your team will use the archive. Useful comparison fields include:

Field	Why it matters
Hot storage price	Cost for frequently accessed media
Cool storage price	Cost for infrequently accessed but online media
Deep archive price	Cost for preservation copies
Retrieval cost	Cost to make cold media accessible
Egress to internet	Cost to download to facilities, users, and vendors
Cross-region transfer	Cost to move media between regions
Minimum storage duration	Penalty risk for recently wrapped projects
Restore time	Whether the tier works for real production requests
API operation cost	Important for large object counts and asset management systems
CDN or partner egress terms	Important for repeated distribution
S3 compatibility	Important for existing archive tools and MAM integrations

The takeaway is simple: compare providers using your restore and delivery pattern. For example, if your archive is a disaster recovery copy that you hope never to touch, a deep archive tier from a major provider may be a strong fit. If you expect monthly reuse, legal pulls, localization requests, and vendor deliveries, a warmer S3-compatible provider with simpler egress pricing may be easier to budget. If your cloud compute, transcode, AI indexing, or finishing pipeline already runs in one hyperscaler, keeping storage near that compute may reduce transfer complexity even if the storage line item isn't lowest.

Model exits before you need one

Egress also affects your flexibility. If it costs a lot to leave, you're less free to change providers, adopt a multi-cloud workflow, or move a catalog closer to a new production partner. This matters for archives because media libraries have long lives. A provider that looks perfect for today’s storage bill may be painful when you need to migrate 500 TB later. Add an exit scenario to the model:

Exit cost =
  total TB to move
x egress rate
+ retrieval fees
+ temporary restore storage
+ transfer tooling
+ validation and checksum labor
+ downtime or limited-access period

You may never run that migration, but pricing it changes the decision. It also helps production leadership understand that “cheap to upload” and “cheap to own” are different. For critical archives, keep metadata and manifests portable. Checksums, file paths, object keys, project identifiers, asset IDs, and storage class information shouldn't live only inside one provider’s console. If you ever need to restore or migrate under pressure, a clean catalog can save days.

Failure modes that create surprise bills

Most ugly egress bills come from ordinary workflow behavior. Your archive may be working exactly as your team designed it. The problem is that your team designed it around access patterns nobody priced. Watch for these failure modes:

A cloud bucket becomes the default source for active editorial downloads
Vendors receive originals when proxies or mezzanines would have worked
Multiple departments independently download the same restored files
Your team stores media in one region while compute or users are in another
Your team uses deep archive for files that need same-day access

Keep the archive accessible, but make access intentional: smaller files for browsing, clear restore paths, visible costs, and defined staging locations.

Put cost signals into the workflow

Give the people requesting restores enough context to make good choices. In a media archive workflow, the interface can show:

File size before restore or download
Storage tier and expected restore time
Whether a proxy or mezzanine already exists
Whether the file has already been staged
Estimated transfer cost for large requests
Region or location of the media
Approval rules for full project restores

Small downloads can stay self-serve. Large restores can route through a producer, post supervisor, or technical director who understands the budget and urgency. In the interface, a 300 GB ProRes master and a 20 MB proxy should feel like different actions. Cost visibility also changes behavior. If users can see that a proxy answers the question instantly while an original restore takes hours and costs money, most will choose the proxy until they truly need the source.

Where cloud archive fits best

Cloud archive is strongest as part of a tiered media strategy. It's a good fit for:

Offsite protection without shipping drives
Distributed teams that need controlled access to archived projects
Libraries where metadata, proxies, and search matter
Burst restore workflows that would be hard to support with only tape

It's a weaker fit when:

Active editorial constantly pulls high-bitrate media
The team can't tolerate restore delays
Outbound delivery volume is high and uncached
Nobody owns cost monitoring or restore approvals
The archive has poor metadata, forcing large speculative restores

LTO, local NAS, SAN, external drives, and cloud are different tools. Many real post environments end up with a hybrid pattern: fast local storage for active work, local or cloud nearline for recent work, cold cloud or LTO for deep archive, and lightweight access copies for discovery.

The decision rule to use with your team

When your team chooses a cloud archive provider or tier, ask the team to estimate three numbers before looking at storage pricing:

How many TB will come out per month or per year?
How quickly do those restores need to be usable?
How many people or partners will request the same files?

When the answer is “almost none, no rush, and only internal recovery,” durable low-cost storage is usually the priority. When the answer is “regularly, same day, and across multiple teams,” low-friction retrieval, controlled egress, and access copies are usually the priority. When the answer is “we don't know,” your safest choice isn't to bury everything in the coldest tier. A warmer layer can hold recently wrapped work while you measure access for a few months, then lifecycle the quiet material down. The best archive lets the right people find, preview, restore, and deliver the right media without turning every request into a budget surprise.

FAQ

Cloud egress is the cost of moving data out of a cloud provider’s network or across certain billing boundaries. In media workflows, it can happen when you download camera originals to a facility, send masters to a vendor, move files to another cloud provider, transfer media between regions, or route data through certain managed services. It matters because large media files can make transfer costs much larger than the monthly storage charge.

Cold storage tiers often have low monthly storage prices, but they may add retrieval fees, egress fees, restore delays, minimum storage durations, early deletion charges, and temporary restored-copy charges. If a project is restored often, used for localization, pulled by VFX or legal, or delivered repeatedly to partners, those access costs can outweigh the savings from cheaper storage.

Model the archive around expected data movement, not just storage capacity. A practical annual model should include storage cost, planned restores, unplanned restore allowance, delivery and distribution cost, provider migration or exit allowance, and operator or engineering time. For media teams, it's useful to model by project type because a scripted series, documentary archive, sports library, and commercial campaign may all have different restore and reuse patterns.

Use smaller access copies for browsing, such as proxies, screeners, thumbnails, transcripts, and searchable metadata. Stage restored files once to a shared location instead of having multiple departments download the same file separately. Keep active and recently wrapped work close to editorial, avoid unnecessary cross-region transfers, cache frequently requested files, and use CDN-like distribution patterns for assets that many external users or partners need.

No-egress or reduced-egress providers may charge more for storage but can be easier to budget when media is downloaded, delivered, or reused regularly. Major hyperscalers may offer very low deep archive storage and strong compute, security, and enterprise integration, but retrieval, egress, operation, and region-transfer fees can change the total cost. The best comparison uses your real restore, delivery, compute, and migration patterns rather than a single price per TB.

Treat the restore as a shared staging event, not a separate download for every editor, producer, vendor, and reviewer. Aspect lets the team work from one shared cloud filespace, so restored media can be accessed in place instead of copied and recopied across departments.