Amazon’s first genuine web service was a source of all kinds of things, from Pinterest and coronavirus vaccinations. In the following years, insiders revealed to Protocol how it developed to hold over 100 trillion items.
In the latter half of 2005, Don Alvarez was just another software entrepreneur trying to get a business idea on the right track when a colleague at Amazon asked him to check out a project that could transform the world.
The company of Alvarez, FilmmakerLive, was designing online collaboration tools for professionals in the creative field and was faced with the same problem that was commonplace at that time storage. Tech startup startups were only beginning to recover from the evil during the dot-com boom, and investing in expensive hardware was not a safe option for startup startups. Purchase too little, and your site will go down. If you buy too much, you’re broke. That was a risky investment in the chaotic world of a startup.
He was still determining what he could learn from movie collaboration with an online retailer. However, he did take the friend’s invitation to collaborate.
“Rudy Valdez blew my brain,” Alvarez informed Protocol. Valdez was the then head of business development at AWS, which at the time only offered the most basic services. He offered Alvarez, now director of engineering at Mural the Mural, a taste of Amazon’s first and perhaps the most important benefit: S3, a cloud-based storage service.
S3, or Simple Storage Support, produced its introduction 15 years ago today. It took a few years before the moment that “the cloud” became one of the most disruptive technologies throughout the past century of computing in the enterprise. Amazon did not even mention the word when it launched S3 the 14th of March in the 14th of March. However, the launch of the storage service immediately solved several difficult issues for entrepreneurs such as Alvarez. It was likely to transform how companies thought about purchasing information technology.
StartupsStartups such as Pinterest, Airbnb, and Stripe were awash with customers to AWS in the next few years. Likewise, more established businesses such as Netflix — which was a DVD-mailing company also made a move to upgrade their processes for the web.
“Amazon was putting infinite disk space in the fingers of every startup at a reduced and pay-for-what-you-need value point. There was nothing like that,” Alvarez said. “The 2nd part was that their API was therefore easy that I could just pick it up and build anything inside in the initial twenty-four hours of using an unreleased, unannounced product.”
S3 has become an essential element of the AWS machine that has generated around $45 billion worth of revenues in the year last. It has grown in numerous directions in the past 15 years. Still, it has kept the same design guidelines formulated by a group led by Allan Vermeulen, Amazon’s chief technical officer, from the very beginning days of AWS, which is at the center of its business strategy.
“We knew what [customers] wanted to accomplish then,” Mai-Lan Tomsen Bukovec, vice president of AWS Storage and the current S3’s head S3, said to Protocol. “But we also knew that programs might evolve since our clients are very progressive, and what they do in most industries will change yearly.”

Building for flexibility
“When people believe greater and quicker in pcs, they think of this,” Vermeulen told me in an interview in 2014 by drawing a line in the air toward the left. But storage technology has developed. Differently, he explained in a time of long plateaus followed by rapid advancements in performance: “It’s the difference between operating my Tesla and traveling my airplane.”
S3 is one of the jarring departures from the norm. It was a blessing for designers like Alvarez, who did not have to think about purchasing and maintaining expensive storage equipment for business purposes.
“There was nothing we had access to that particular provided anything slightly like what S3 can do,” Alvarez declared. “I felt like somebody had just provided me the recommendations to the candy store.”
As with the rest of AWS, S3 was born from the experience of Amazon in building and expanding Amazon.com and gave it many difficult lessons about the limitations and the possibilities of spread computing.
“A forcing function for the design was that the single Amazon S3 spread program must support the needs of equally inner Amazon purposes and additional application designers. This implies it must be fast and reliable enough to run Amazon.com’s websites. At the same time, variable enough that any developer can use it for almost any information storage requirement,” AWS stated in its initial press release to announce the launch of S3 in 2006.
At the beginning of cloud computing, reliability and performance were major concerns. These concerns were particularly sensitive to data, which, just 15 years ago, was thought to be among the most valuable assets a company could have.
“Once we introduced S3 15 years back, S3 had nine micro services, and we have more than 300 now.” Tomsen Bukovec discussed the new software development method of breaking large, interdependent portions into smaller, separate services.
The microservices model helped AWS to centralize points of failure in S3 and create an infrastructure that recognizes that cloud-based services distributed may occasionally fail and that failures should not bring the whole system down.
It also allowed the company to add future improvements without disrupting the fundamental components of the system. AWS is now claiming that S3 has “11 9s” of reliability and an impressive 99.999999999 percent uptime, which is higher than self-managed storage devices by a substantial margin. (Other cloud storage providers have also surpassed this benchmark.)
S3 started as an archive pen for basic web elements such as videos and images that web operators would upload through AWS and then upload to the browser whenever you opened a page. As companies were more familiar with using the cloud, they began storing all kinds of data in S3.
That’s when things started to get messy.

Fixing leaky buckets
If you go back to the various security incidents in the last few years, the majority could be traced to “leaky buckets,” referring to the primary component in S3 storage. The same happens with other cloud companies, too; however, given the size of the market share of AWS, it’s an issue that AWS has faced in a variety of instances.
AWS operates in the “shared responsibility” design for security: AWS can prevent anyone from accessing the servers or infiltrating its network. However, customers are expected to secure their accounts to a certain extent. That is, it’s not your fault if you’re a rental car company if someone can steal your laptop from the back of an unlocked car.
But time and time, cloud users have left sensitive information belonging to their customers in storage buckets that are not secured and accessible to anyone who wants to locate them, which is much easier than you think. This is only one instance of how AWS must adapt certain of its core services to serve customers no matter where they are. This is especially true for those who have arrived later and are accustomed to accessing their data via private networks.
“In a business application earth, you don’t need accessibility external the organization or a small grouping of consumers within the business,” Tomsen Bukovec stated. However, it was evident that AWS required more to assist its customers in aiding themselves, which resulted in the creation of tools like Block Public Access that could secure every storage bucket associated with corporate accounts.
It was also apparent to anyone outside the beginnings of AWS that Amazon’s famed “two-pizza teams” were “equally a power and a weakness,” Alvarez stated.
“It enabled all those services to rocket forward at a speed none rivals can match. And in the early times, there was less consistency [and] which was hard to solve and handle,” He explained, adding that the process has improved with time.
Further security tools have also been introduced, which allow customers to scan their accounts for access to the internet or give different access levels to individuals who hold various positions within a company.
“Wherever we’re viewing customers move making use of their migrations is which they frequently have hundreds of containers and plenty and lots of [different] roles,” Tomsen Bukovec claimed of those new to the cloud who are the most likely to make making these mistakes. “Once we think of what to build to help consumers secure the border of the AWS source, we think of how they want to audit and how they want to control” accessibility to their storage assets in S3.

The goal is to reach 100 trillion.
S3 was constantly evolving in the years after its launch as well as becoming much more affordable: In the year that AWS was able to hold its first major re: Invent developer event in 2012, One of the most significant announcements that took place during the conference was an increase of 24% to 28 percent reduction in S3 storage costs, which was the 24th time such price cuts AWS had announced in the past.
The price reductions were made feasible because AWS could modify the base S3 service at any time, according to Alyssa Henry, the then vice-president of AWS Storage Services. Vice president of AWS Storage Services stated in a keynote address in 2012.
S3 was initially designed to store 20 billion objects. However, it grew faster than expected, reaching 9 billion things in its first year. The company upgraded the base storage service to accommodate more capacity in mind with no interruption to the initial S3 customers. by 2012, it had grown up to hold 1 trillion items in storage and, in 2020, 100 trillion.
“What’s cool about this is customers didn’t have to do any such thing: You didn’t need to head out get another upgrade — v2 of Amazon S3; you didn’t have to migrate yourself; you only first got it all for free. It worked. Things just improved,” Henry, who’s today government vice-president and the head of Square’s Seller division in 2012, told the conference. “That’s one of many variations with the cloud versus how old-fashioned IT has been done.”
An identical upgrade was released in the last few months, as AWS established a strong uniformity across S3.
Consistency is a notion of data storage that can jolt your brain when it first appears. Earlier storage systems like the initial S3 were built to support “eventual consistency,” meaning that a storage system could not always instantly tell whether a new bit of data had gotten in its storage bucket. Still, it could catch up in the nick of time.
Nowadays, modern applications run much faster. Any application that sends an inquiry to the storage service requires up-to-date information available to run at the highest level. Over the past few years, AWS revamped S3 with strong, consistent principles that other cloud providers provide but could launch against fewer users.
“That is an extremely complex design issue,” Tomsen Bukovec declared in one of the most memorable announcements at the re: Invent 2020 with the geekier crowd of AWS users.
As they enter the new decade, Tomsen Bukovec and her team are looking into ways that will make it simpler to apply machine learning to the over S3 data and also to increase the efficiency and performance for data lakes, which enables an analysis that is more precise of customer and internal data in AWS users.
The Moderna vaccine against COVID-19 was created with the aid of an S3 Data Lake, according to Tomsen Bukovec.
“We have this unique view that individuals accumulated more than 15 decades of use, wherever we could know what our customers are trying to do and how we could construct [S3] in this way, so it keeps true to that particular simple, cost-effective, protected, sturdy, trusted and highly-performant storage,” she added.