 Live from Las Vegas, Nevada, it's theCUBE. Covering AWS re-invent 2016. Brought to you by AWS and its ecosystem partners. Now here's your host, Stu Miniman. And welcome back. You're watching SiliconANGLE Media's theCUBE here at AWS re-invent 2016. It's a two and a half days worth of interviews and of course we can't end our coverage without bringing back one of our favorite guests from last year, my lawn, Thomson Bukovac, who's the Vice President General Manager of Amazon S3. Thank you so much for joining us. Stu, thank you for having me back. Yeah, absolutely. So, you know, super exciting show of so many people here at the event. What was it? I think 14 main announcements and your group had a lot going on. So I guess let's start with, you know, since we talked last year, kind of the main updates, what's happening in your business? Well, the way I think about it, Stu, is that this was the year of file. We have some great services for block storage and for object storage with S3 and Glacier. This is the year that we introduced not one, but three new capabilities for file storage. And as you know, there are many companies out there who have a lot of investments in applications and in storage that are file servers. So now we have these three options. We have Elastic File Service, EFS, and that is a fully managed file service. One of the things that we announced just this week was the ability to connect EFS with an on-premises environment using our direct connect service so that you can just keep your investment and your applications on-premises and use our fully managed, highly durable, highly available EFS for file storage. We announced the file gateway for S3, super common request. Customers have said, how can I get files from my on-premises environment up into S3 just as an object in a bucket? So any application, whether it's on-premises or an EC2 instance, can go access that object for data processing or any other type of operation. We have that capability today and it's really easy to set up. And the third, as you know, is our new version of Snowball, which I think you're going to like a lot. Mylon, you talked to the storage people out there. I mean, you guys, I think it's Amazon, it's like object. Storage engineering, object, it's about time. Object was really built for this. What you built, it's not your, you know, it's not your father's file system. Sounds like you're doing some really cool things. What drove the creation of those solutions? You know, was this customer-driven, you know, something to fill out the portfolio? It's not a checkbox from an Amazon standpoint. So you can understand how this is different from, you know, what we've been doing files for the last couple of decades. Well, I think the way we think about storage absolutely comes from what customers are asking us for. And if I think about the different things that we've developed, not just on the file side, but also for Glacier, for EBS, and for S3, it's all driven by customer request. So I talked a lot about how customers have all this file storage today and they want to bring it to the cloud. They want the burst ability. They want the elasticity and the scaling of it. But they also want to have their storage in different areas operate in different ways. So I'll give you an example. S3 has object storage. And what customers have told us is that they want to make S3 a hub of all the different types of storage they have. So this week we introduced the ability to put a customer defined tag up to 10 on an individual object and that tag can be edited at any time. So you can put the name of your application on that tag. You can put a custom ID. You can put a compliance attribute on that tag. And then in S3, you can tear automatically to a different storage option like Glacier. You can look at operational metrics viewed by that application name and a tag. Or you can even set access control on whatever value is in that tag. And we're coming up with this, not because we're in some ivory tower and we think we want to make super smart platforms. We're coming up with this because customers say, hey, I have all this business logic and I want to use it. And I want to use it with Lambda functions and I want to use it with your tools. And we do that all the time. For example, for EBS, customers love EBS. It's super low latency storage. It's block storage in the cloud that's fully managed. But they want more options. So this year we introduced high throughput EBS volumes and we have cold EBS volumes. And we now have these tiers, these different tiers for EBS volumes, customers can just pick which ones they want. And that choice is driven by our conversations with them and us going back and seeing how we can do it. All right, so file, it talked about block, bread and butter, S3. Sometimes it gets overlooked a little bit that we know S3, it's really the de facto standard in the industry for object today, driving a lot of innovation across the industry, what kind of leadership there. But how's today's S3 different from what people might known if they were trying it a year or two ago? Well, Stu, S3 has been around for 10 years now, 10 years. We have customers who have exabytes of data. And so one of the things that we look at is A, just continuing to drive the platform. How can we make it better? We introduced native IPv6 support. We continue to add things like tags. We recently added just this week an inventory of buckets. You can just turn it on and every day you get a list of what's in your bucket. But I think one of the most exciting areas for S3 is around analytics. So this is the year that we have introduced a feature that if when you turn it on, we'll start observing the access patterns for storage in your bucket. And then after we collect a set of observations, we will show you the heat, the patterns of access on your bucket so you can make cost decisions. You can go have conversations with data. We run our business with data and we think customers do too. They tell us they want that data to make the type of decisions that they need to on storage. So when we collect these observations, we display it in our console, but we also let you have a comma delineated text file that can be opened up in Excel for the pivot table geeks out there and I know there's a lot. Open up a quick site or any business analytics tool. So the direction I see S3 going is continuing to be the absolute best of class storage for anything you want to do, whether it's big data or archive or media storage, but we also see it continuing to develop a whole platform of storage management capabilities because we want customers to have that insight into their storage. Yeah, it's really interesting. The storage industry, it's about, no longer just storing data, it's how do I get value out of my data? And I hear it every year here at the show, but I felt like it's finally starting to sink in as to talk about analytics, IOT a lot more as to how those people, solutions come in, serverless as you talked about, trying to pull into it. Maybe in hybrid is an update there. Maybe we talk about hybrid, what that means now in your storage environment. Well, customers have been telling us that they want to have different ways to integrate with their on-premises environment, and Stu, this is nothing new. S3's been around for 10 years, AWS has been around for 10 years, and since the beginning of our services, we've always worked with enterprises, and there are many enterprises who aren't native cloud. So this concept of hybrid is not new to us. I think what's interesting about now and how we're integrating with those on-premises environments is that we've really, we've really come up with some innovative ways to do that. So the File Gateway is one example where you can have a virtual appliance on-premises that sends data back up, but we've come up with some new ways also. And one of the ways that we've come up with integration for on-premises is our new Snowball. Yeah, so Snowball, it's really interesting. You know, when you walk around the show, people are really excited. Andy Jassy walks by, everybody's snapping photos. They talk to James Hamilton, and he said, oh, people want to take selfies. They want autographs. You brought along a special guest that everybody's taken selfies with. So at the end of the keynote, a truck came out, you know, talk about the snowmobile. Unfortunately, we didn't have room on the set here for the full truck, but maybe you can introduce our special guest. Well, Stu, here I have a Snowball Edge. Now, I did try for the truck, Stu, but I wasn't able to wheel it up, so I have the Edge instead. And what you see here is a smaller form factor than the original Snowball. It's a little less. It's about, it's a little under 50 pounds. But what's inside this is actually kind of revolutionary. Inside this device is compute. And so what we've done is we've essentially put a file server and an S3 endpoint and the compute power of an M404 XL. We also have 100 terabytes of capacity in storage here. And the way that we've built this, if you think about that first Snowball, that first Snowball was all about data transport. It was about getting storage from point A to point B. This is a little different. This is an Edge device. And we expect that customers are going to keep it on site for a little more. And they want to keep it on site for a little bit more time because they want to do data collection. So you can cluster this. You can put up to five of these Snowball Edges in a cluster and you can keep it for a while. It's the same cost structure where you pay for the device plus shipping and then you pay a daily fee. So this device here costs $300 plus shipping. You get the first 10 days on site free and then after that you pay $30 a day. And you can keep it as long as you want. So we have customers like Philips. And if you ever want to talk about a constrained data center, a hospital is a constrained data center. They have to keep all their servers in places like that basement. And so what hospitals are doing with Philips is they're collecting off instruments, MRI data, and they're collecting it on these Snowballs. And because we have the compute power in the Snowball, it's doing transcoding so that when a medical professional wants to take the MR data out, they can take it out and transcode it on the fly. Now the hospital may or may not send it back to Amazon. They might send it back if they want to keep a copy up in the cloud in S3 where they might just keep it there and treat it for bursting capability if they need to. You spoke, it's an interesting use cases. IOT, CM01, how do I get to all of those centers where I might have no connectivity? I think you said you could put this thing on a boat, go collect all your data, and then it's got lots of connectivity. It's got options. And you said it's got compute. It doesn't just have compute. It's got serverless, lambda, green grass. Explain why that's pretty important to have that kind of outside of the Amazon cloud. Right, so this support for lambda taps into the serverless computing capabilities. And so what people are able to do is they're able to write simple lambda functions that look for anomalies in the data that are coming in, very important in the IOT space. You have the sensors streaming all this data in from windmills or any other IOT device and a lambda function can do parsing of that and it can look for anomalies and alarm on it. There are so many different things that can happen with this. We have Oregon State University who does put it on a boat. They put it on a boat because Oregon State goes out and they collect oceanographic information and they want to bring it, they want to do some processing of that information on the boat and the type of processing that you can do with a lambda function ranges from deduplication of data to actually looking at the data and processing it in some way. It opens up all these different opportunities for developers to use a device for processing on the edge with a whole lot of storage. All right. Mylon, I want to give you the final word. We talked last year, just innovation in the storage industry was something that people kind of bemoaned, kind of the lack of innovation for quite a while. As you look forward, what are you excited about? We come to 2017, maybe think about what kind of things you and the storage industry are tackling that we'll be able to handle. In AWS, we don't necessarily think storage and compute. We think what do customers want to do? And I'll tell you, Stu, that truck, the snowmobile, is something that we came up with because customers told us I want to figure out how to move exabytes of storage. And so we built a truck. We built a truck that has a 45 foot container, 100 petabytes of capacity. It's designed for up to a terabit of connectivity. And we built this, not because it's a theoretical thing, we built this because we had customers like Digital Globe that collect geospatial data from satellites and they generate over 85 terabytes of data a day. They came to us and said, how can we move all of this data into the cloud? So we don't think about storage, actually. And that might be funny coming from me and us three. We think about what do customers need to do with data and how can we go solve that problem? And I think that's what's so exciting for us. We don't have these boundaries that we think about. If we need to build a truck, we build a truck. We need to build a regularized form factor of 50 pounds to go sit and some edge location, we'll build that. We'll do anything that customers ask for and give them places to put their storage. Mylon, great to catch up with you. Thanks so much for bringing the snowball on to give us the update here and congratulations on all the progress and everything. We look forward to talking with you more in the future. That's wrapping up our coverage here of AWS 2016 for Stu Miniman, John Furrier, the whole SiliconANGA media crew. Thanks so much for watching theCUBE and look forward to seeing you at many more events.