🪣 An Ode to Amazon S3
Some technologies shout about their greatness. Others just… store the internet.
Amazon S3 belongs firmly in the second category.
Launched on 14-March-2006 with a simple promise — store anything, retrieve it anytime — S3 has spent the last two decades becoming the unsung hero of the cloud. Photos, backups, data lakes, ML datasets, logs, websites… if it lives in the cloud, there’s a decent chance it lives in an S3 bucket.
And like any 20-year journey, the story of S3 is filled with innovation, lessons learned, and a few “well, that was interesting” moments.
20 Years of Quietly Running the Internet.
Let’s take a nostalgic walk through the milestones:
🚀 2006 – Amazon S3 Launch
Announced as "storage for the internet", designed to make web-scale computing easier for developers, it supported REST, SOAP, and BitTorrent protocols. Storing 1 GB of data for 1 month costed just 15 cents, while transferring data in-and-out of the system costed 20 cents per GB!
🌍 2007 – S3 in Europe
S3 expanded outside US, with its first region Europe, reducing latency and making cloud storage truly global. For European developers, this meant their data didn’t have to take a transatlantic vacation every time an application accessed it.
⚠️ 2008 – The Famous S3 Outage
While the announcement of Amazon S3 storage price reduction attracted more customers, a major outage briefly reminded everyone that even cloud infrastructure can have bad days!
🌐 2009 – CloudFront Integration
Integration with Amazon CloudFront allowed S3 to power global content delivery. Suddenly S3 wasn’t just storage—it became the engine behind faster websites, media streaming, and global apps. Requests originating anywhere in the world were routed to one of 14 edge locations (8 in the United States, 4 in Europe, and 2 in Asia). By the end of 2009, Amazon CloudFront that the ability to limit access to Amazon S3 content using Origin Access Identity (OAI).
💰 2010 – Reduced Redundancy Storage
The Standard Storage Class now provided 99.999999999% durability.
AWS introduced Reduced Redundancy Storage for data that didn’t need the full durability treatment and provided 99.99% durability.
In simple terms: “Cheaper storage for stuff you wouldn’t cry about losing.”
📦 2010 – Multipart Upload
Multipart upload allowed large files to be uploaded in pieces simultaneously.
Anyone who has ever tried uploading a massive dataset over a flaky connection knows how revolutionary this felt—finally, uploads that don’t restart from zero.
2010 brought with it a lot of features still available to us - support for bucket policies to set access control for buckets. S3 accessible via the AWS Management Console, AWS Cloud Free Tier allowing storing up to 5GB of data free of charge, but most importantly - support for storing large objects up to 5TB in size.
🌎 2011 – Static Website Hosting
S3 suddenly gained the ability to host static websites directly from buckets. Developers realized they could run websites without servers, and a whole generation of lightweight web apps was born.
🔐 2011 - AWS Identity and Access Management (IAM)
IAM was Generally Available, and management features of Amazon S3 are officially available
⚙️ 2011 – Lifecycle Policies
Lifecycle rules automated the movement or deletion of data based on age or usage.
For engineers drowning in storage bills, this felt like hiring a very responsible robot janitor for your buckets.
🧊 2012 – Amazon S3 Glacier
The arrival of Amazon S3 Glacier brought ultra-cheap archival storage to the cloud, without compromising the durability!
Companies that once stored tapes in warehouses suddenly realized they could archive decades of data without needing forklifts.
🌍 2012 – Cross-Origin Resource Sharing (CORS)
CORS support allowed web applications from different domains to safely fetch resources from S3. It quietly enabled modern web architectures where APIs, apps, and assets live in different places but work together seamlessly.
🧾 2013 – CloudTrail Logs to S3
The announcement of AWS CloudTrail, a web service that could records API calls made on your account and delivers log files to your Amazon S3 bucket. Just like that, S3 became the default vault for security logs, audits, and compliance data.
🗂 2014 – Lifecycle Management of Versioned Objects
S3’s Lifecycle Management integrated S3 and Glacier and made the details visible via the Storage Class of each object. You could set up a simple Lifecycle rule using the AWS Management Console.
🔔 2014 – Event Notifications
Support for event notifications to trigger events to Amazon SNS, Amazon SQS, AWS Lambda - whenever objects were created or modified. Suddenly storage became interactive, kicking off serverless pipelines and automation workflows.
🔎 2014 – Server-Side Encryption with Customer Keys
SSE-C allowed customers to supply their own encryption keys. Security teams everywhere collectively nodded and said: “Now we’re talking.”
🔐 2015 – S3 VPC Endpoint
VPC endpoints allowed private connections between S3 and workloads inside a VPC. Data could now move without touching the public internet, which made security architects sleep noticeably better at night.
2015 also brought along another key feature - cross-region replication
⚡ 2016 – Transfer Acceleration
Transfer Acceleration used AWS’s global edge network to speed up uploads. If you’ve ever uploaded huge files across continents, you know this feature felt like someone secretly upgraded your internet connection.
🧠 2017 – Amazon Macie for S3
Amazon Macie began scanning S3 buckets for sensitive data using machine learning. Finally, a system that could say: “Hey… maybe storing credit card numbers in that bucket wasn’t the best idea.”
🛡 2018 – Block Public Access
Block Public Access was introduced to prevent accidental exposure of buckets. Let’s just say this feature has saved countless engineers from awkward security incidents.
🧠 2018 – Intelligent-Tiering
Intelligent-Tiering automatically moves objects between storage tiers based on usage patterns.
Translation: S3 started optimizing your storage bill while you slept.
📊 2018 – One Zone IA
One Zone Infrequent Access offered lower-cost storage in a single availability zone. Perfect for secondary backups or data that doesn’t need the full multi-AZ treatment.
🎯 2019 – S3 Access Points
Access Points simplified permission management for large shared datasets. For organizations with thousands of users and applications, this was like finally organizing a chaotic storage closet.
⚡ 2020 – Strong Consistency
S3 introduced strong read-after-write consistency across all regions. For developers, this was a huge moment—one of the most annoying distributed system edge cases simply disappeared.
🌎 2021 – Multi-Region Access Points
Multi-Region Access Points allowed applications to access replicated data across regions through a single endpoint. Global applications suddenly became easier to build—and far more resilient to regional outages.
💾 2022 – AWS Backup for S3
Integration with AWS Backup allowed centralized backup policies for S3. Because when it comes to data, the only thing better than storage… is having a backup of that storage.
🚀 2023 – S3 Express One Zone
S3 Express One Zone introduced ultra-low latency storage for high-performance workloads. For AI pipelines and analytics systems, this meant faster data access and happier GPUs.
🤖 2024 – Amazon S3 Tables
S3 Tables introduced native tabular storage optimized for analytics workloads. This pushed S3 deeper into the world of modern data lakes and large-scale analytics platforms.
📦 2025 – Maximum Object Size Increased to 50 TB
A major announcement at re:Invent 2025 - the maximum object size increased to 50 TB. At this point, uploading a single object to S3 could basically mean “store an entire dataset in one go.”
🌐 2026 – Account Regional Namespaces
Just a day before the 20th birthday, the announcement dropped like a perfect pain patch — removing the age-old friction of checking if your bucket name is already taken before you can start storing data.
With this feature, you can predictably name and create general purpose buckets in your own account regional namespace by appending your account’s unique suffix in your requested bucket name, meaning you can now create general purpose bucket names across multiple AWS Regions with assurance that your desired bucket names will always be available for you to use!
Conclusion - ☁️ The Service That Quietly Holds the Internet
Amazon S3 rarely gets flashy headlines. It doesn’t launch new phones or social networks.
But behind the scenes, it stores trillions of objects and exabytes of data powering applications across the world.
From startups to global enterprises, from static websites to AI training datasets—S3 has been the quiet backbone of the cloud revolution.
And after 20 years, one thing is pretty clear:
The internet may run on code… but a lot of that code runs on S3.
Which S3 feature do you think changed cloud architecture the most? Drop your thoughts in the comment section!
Bonus
Checkout the original S3 release announcement and Jeff Barr's blogpost from 2006!
https://aws.amazon.com/about-aws/whats-new/2006/03/13/announcing-amazon-s3---simple-storage-service/
Top comments (0)