AWS' Kevin Miller talks S3 in 2023 and beyond
Amazon S3 has become a massive portfolio of products, which only grew at re:Invent 2022. Here, the vice president and general manager of S3 talks about what's next.
LAS VEGAS -- At re:Invent 2022, AWS unveiled incremental updates and improvements to its storage portfolio that focused on speed and data access, as well as support for customers with exabyte-size storage needs.
But in this Q&A, Kevin Miller, vice president and general manager of Amazon S3, made a point of focusing on a new capability beyond storage technology -- AWS Key Management Service (KMS) External Key Store for key encryption -- as important to the object storage service's future. He also provided a storage take on Amazon DataZone, a new data management service.
TechTarget Editorial spoke with Miller about the ongoing evolution of S3 storage, potential changes to the storage portfolio, and what future services AWS is developing to corral ever-rising storage amounts and costs.
How is S3 going to continue to evolve for Amazon?
Kevin Miller: I want S3 to be a storage system that partners -- both first-party and third-party services -- can use on equal footing.
For 2023, we're increasingly working with customers that are building large-scale data lakes using S3. [AWS CEO] Adam Selipsky talked about how Pinterest now has more than an exabyte of data. That's a great example of a customer doing a lot with AWS -- collecting a lot of data and driving a lot of business value out of that data.
That's a theme that a lot of customers talk about to me: what more [they can] do with data, ideally, to drive top-line growth, but other benefits as well, whether it's cost optimization or developer agility.
What will be the next major innovation in storage for AWS over the next few years?
Miller: A few years ago, we launched the idea of access points, and last year, we launched the idea of multi-region access points. Those are ways that customers can create separate security policies for particular applications that want to access data in S3. Multi-region access points provide multi-region resiliency for customers who have critical applications that need failover operations into a different region.
Another thing I would highlight is the innovation and the iteration around encryption. We now support three different ways of doing server-side encryption. Today, with the announcement around KMS, that's a capability that S3 customers can use. The API hasn't changed dramatically since we launched it [in 2006], but we've added that ability to have encryption on top of it.
The thing we're going to do a lot more and continue to do is helping and investing in open source connection software that uses S3. We recently announced new contributions we made to the Trino open source package.
We have developed a new software called ShardStore, a completely new storage layer at the very base of S3. [It's] new software we've used some advanced techniques around testing to prove the correctness of.
Are there any plans to streamline or cut prices on storage in 2023?
Miller: We're pretty happy with the portfolio we have, particularly with Intelligent-Tiering. That's the storage class we recommend as a default for most customers because it automates the shift of data into those lower-cost access tiers. We have the other storage classes for specific workflows.
Kevin MillerVice president and general manager, AWS S3
The new Amazon DataZone unveiled at re:Invent appears to go after data management vendors, a vertical that has become aggressive in marketing to storage administrators in recent years as the role has evolved. What niche will this native service fill that similar vendors have not?
Miller: DataZone is a critical capability for many customers looking for an improved ability to discover and share their data across the organization. [They then] have all the governance, controls and visibility into how that data is being used.
[Customers] are looking at what more can they do with data and how to embed data into business decision-making. [They want to] move away from, in many cases, data in storage being something that just a central IT team focuses on -- how [instead can they] use those data sets to drive top-line growth or optimize [their] manufacturing process.
Editor's note: This interview has been edited for clarity and length.
Tim McCarthy is a journalist living on the North Shore of Massachusetts. He covers cloud and data storage news.