Skip to content

Instantly share code, notes, and snippets.

@davilima6
Created June 10, 2026 11:34
Show Gist options
  • Select an option

  • Save davilima6/3943567a6592102e0646d42b5904c4b0 to your computer and use it in GitHub Desktop.

Select an option

Save davilima6/3943567a6592102e0646d42b5904c4b0 to your computer and use it in GitHub Desktop.
Amazon S3 Glacier Deep Archive

the s3 cheapest storage uses tapes or some ancient storage device, very slow?

You actually hit the nail right on the head! The absolute cheapest tier of Amazon S3—S3 Glacier Deep Archive—is the cloud equivalent of a high-tech digital warehouse. It relies heavily on a technology many assume is obsolete: magnetic tape (alongside heavily customized, powered-down hard drives).

While Amazon keeps its exact hardware proprietary, industry teardowns and system behaviors point to massive, automated robotic tape libraries handling this data. Far from a dusty 1980s cassette player, these are cutting-edge, ultra-high-density enterprise tape systems.

Here is exactly how it works, why it is so cheap, and just how slow it can be.


Why Use "Ancient" Tape?

For long-term storage where data is rarely touched (like compliance records or old backups), magnetic tape is still the undefeated champion of data centers for three major reasons:

  • Cost: It costs a fraction of a cent—roughly $0.00099 per GB/month (about $1 per Terabyte).
  • Power consumption: Spinning hard drives or SSDs require constant electrical power and cooling. A tape cartridge sitting on a shelf consumes zero power until a robot grabs it.
  • Durability: Tapes don't suffer from the mechanical or electrical failures that active hard drives do over a decade.

Just How Slow Is It?

It is incredibly slow compared to regular cloud storage. While standard Amazon S3 delivers your files in milliseconds, getting data out of Glacier Deep Archive takes 12 to 48 hours.

This delay happens because the data is essentially "offline." When you request a file, AWS kicks off a multi-step process called rehydration:

  1. The Queue: Your request goes into a massive internal job queue.
  2. The Robot: A literal robotic arm in an AWS data center navigates to the correct shelf, grabs the specific tape cartridge containing your data, and slots it into a tape drive.
  3. The Sequential Scan: Unlike an SSD, which instantly jumps to a file, tape must be physically wound forward sequentially until it reaches your data.
  4. The Transfer: Your data is read, decrypted, decompressed, and copied back into a "warm," fast-access S3 bucket where you can finally download it.

S3 Storage: Fast vs. Deep Archive

Feature S3 Standard (Warm/Hot) S3 Glacier Deep Archive (Cold)
Primary Media Fast Solid-State Drives (SSDs) & Active HDDs Robotic Tape Libraries & Spun-down HDDs
Retrieval Speed Milliseconds 12 to 48 hours
Approx. Cost ~$23.00 per Terabyte / month ~$1.00 per Terabyte / month
Best Used For Active websites, apps, and daily data Emergency disaster recovery, 10-year regulatory audits
Retrieval Cost Free to pull data out You are charged a fee per GB retrieved

⚠️ The Golden Rule of Deep Archive: It is a fantastic tool to slash your cloud bill, but you should only put data there if you are 100% certain you won't need it in a hurry. If your business depends on getting that file back in under a day, Glacier Deep Archive will turn an emergency into a waiting game.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment