AWS Re:invent in review - CTO045

Machine-generated from the episode audio and not hand-corrected, so names and technical terms may be imperfect. The audio is authoritative.

All right, welcome to the CTO Advisor. You know what, it's been a while. It's been at least a month and a half since I've done a podcast episode, but that's okay because I have a really special guest. It's going to be worth the wait. We're going to talk a little bit of AWS reInvent, how AWS is attacking legacy infrastructure providers and how Stu, who's going to talk me off the ledge, how I completely believe that AWS is going full tilt into hyper-converged infrastructure and Stu's going to tell me why I'm wrong, but I think we're going to get to a point where we both get to the same spot.

For Stu, for the two people out there that don't know you, can you go ahead and introduce yourself? Hey Keith, great to talk with you again. Hi audience here, Stu Miniman, at Stu on Twitter and I'm an analyst at Wikibon and a host of theCUBE video program fresh off of the Amazon reInvent show. And yeah, I mean, Keith, last time we talked on the podcast was when we were getting ready for VMworld. So for me, always VMworld and Amazon reInvent, two of the biggest shows in our industry and excited to dig into it with you.

Yeah, 32,000 people, what was that like? So Keith, how many times have you been to the Sands, you know, Venetian Palazzo area? A lot, more than I would like to admit. Let me, I've lost track of how many times I've been there and I was blown away. I got there on Monday when they were doing some kind of pre-show analyst activity and they had some of the partner stuff and already kind of two days before the main event started, it was crawling with more people than I'd ever seen at that facility.

When they actually got to the keynotes in the main part of the event, I mean, it was just a swarm of people. You know, the cloud arrived in force. They actually spread out to a couple of other facilities but it was massive. You know, logistically, like trying to get into certain sessions or just get through the hallway at certain times, the show were really brutal and the way they're growing, I think they're going to have 50,000 people at the show next year.

So, you know, everybody's talking about how that's even going to be possible. Yeah, they might have to actually head to the convention center of all places outside of one of these traditional hotel events. You know, Vegas is huge and they're used to accommodating these, I'll say, smaller tech conferences of 20,000 people, 20, you know, two to 25,000 people for VMworld but 32,000 people for AWS reInvent. I really hate that I missed it. I really wanted to go this year and I think this year was, you know what, the what is the fourth or fifth year now?

The fifth year of the show and it was my fourth year going and yeah, I mean, just to your point, Keith, we're going to be comparing the show to CES, not the rest of the enterprise shows going forward. So I think it's with that backdrop, I don't think we could probably overhype just volume that AWS is taking out of the room when it comes to conversation with other enterprise vendors. The talk last week was all about AWS. I made this wild statement that AWS is going to compete against Nutanix.

We'll get into that in a minute but did you hear any other news last week outside of reInvent? Well, I was so busy at the show, I always call my wife when I get to the end of a three-day show like this and say like, just let me know if a meteor hit or anything else major happened because, you know, you're kind of in this bubble when you're at the show. So no, it was pretty tough for anything to dent that and it was such a busy event.

I literally had a day that like, I didn't get to check email at all just because, I mean, you know how it is, Keith, you wake up, you're in breakfast meetings all the way through the day, through the evening and just nonstop and so many people that were at the event that you want to try to get a few minutes to talk to. So yeah, it definitely, being inside, there could have been some amazing stuff that happened out in the world but it was pretty rare that anything made a dent.

Yeah, so one of the great announcements or more interesting announcements that came out of the show was the Greengrass platform, this Edge IoT computing platform that AWS announced and they announced it in conjunction with the Snowball Edge, a clusterable update to the Snowball which allows you to run some Greengrass compute on it. It's a hybrid type of solution. In my world, I looked at it, I'm like, that's a competitive, I want to use my words carefully, competitive in the AWS sense, how AWS competes with legacy enterprises but if you had a use case for putting compute and storage to get together at an Edge location, this was kind of Amazon's answer to that HCI use case.

How did you see it? Yeah, so I'll give you my quick thoughts on this, Keith. First, from an architect standpoint, I understand what you're saying. When I want HCI, I've got compute, I've got storage, I've got some ability to cluster it together, it's got some networking built in, it looks kind of interesting. What I want to do is, if you look at the use case, actually there was a good article written by the Redmonk folks looking at the Snowmobile which is part of that Snowball family.

So you had the Snowball which was announced last year which is just storage, put it in the box and then ship it back to Amazon. You've got the Edge which, as you said, has some of that really cool Lambda stuff, serverless, to be able to do something on the box and then you have the Snowmobile which is this truck that fits 100 petabytes worth of data. And by the way, quick aside there, Amazon on the Snowball Edge which is 100 terabyte box says that it is quote petabyte scale and the 100 petabyte truck does quote exabyte scale.

So I think Amazon's got to figure out an extra order of magnitude for both of these things but it's interesting. When they looked at the Snowmobile, it's not like it's a data center that pulls up in a truck that runs. It's really about loading data into it and then eventually when you're done, you ship it to the data center. And if I look at the Edge, here's the use case. This is not general purpose compute. This is not, you're not going to run your VMware on here, you're not running VDI, you're not running any Microsoft apps on here.

Yes, it has data and I can run compute on it but really the first application that they're using this for is collection of data at the Edge and Snowball Edge and especially like these analytics types things pull a lot of information off of sensors especially where they might not have any or very little connectivity. So think of sticking it on a boat, putting it out by the windmill, somewhere where I can collect a bunch of data. I want to have a few of them so I've got redundancy and eventually I'm going to ship some or all of that back to the public cloud.

So it still is kind of this temporary box. Yeah, it has some of the same kind of checkbox features that I might expect from kind of a clustered architecture but compared to HCI which is a platform for my kind of on-premises private cloud if you will type of environment, it's very different use case, very different job it's getting done and I'd be shocked if I ever found somebody debating vSAN versus Nutanix versus AWS at that level. Yeah, I think I absolutely agree with you on most points.

You're not going to get into a debate of vSAN versus Nutanix and Snowball Edge. I think what Snowball Edge represents is a major departure from AWS in the sense that the use case that Jassy gave during the keynote was this continuous operation mode where you'd have a Snowball Edge at a wind turbine collecting sensor data and then having simple Lambda functions running on it to do some real-time analysis so you can do alerting because there's this use case where you need some compute as it's collecting data so that if a sensor goes out of range it can alert whatever operations team or kick off a manual or automated process to rectify an issue, you need that capability as well.

So Snowball Edge, what surprised me was that Amazon was willing to operationalize compute at the edge outside of their boundaries, outside of kind of their safety zone was one of the more interesting parts of that solution. Yeah, and Keith, I totally agree. Last year when Amazon announced their IoT solution, my CTO David Floyer wrote some articles and we had a few conversations with Amazon because just the physics of moving data is really tough. So while we're not arguing that there's general shift towards public cloud and things like the original Snowball make it easier for me to move data and we might build new applications, if you look at IoT, the vast majority of the data we think it's like over 90% of data for that kind of solution is going to be created at the edge and most of that data is not going, it doesn't make sense to bring it back to that central hub.

Even if I'm doing my own cloud, it's like I need to be able to send the compute to where the data is. It's something we've been talking about in kind of the big data space for a number of years and this is Amazon saying architecturally, yeah, this makes sense, this is why we're creating this solution, we want to enable customers to use this and it's really an extension. When they put that Lambda stuff in there, I called it, it's like a mini AWS that you get to put at the edge and really extends what they're doing in the cloud.

Which makes it a hybrid solution and it's interesting Amazon's take on what hybrid is. Like what the VMware thing is, the AWS, I mean the VMware cloud and AWS and now this, how do you interpret Amazon's take on hybrid? Yeah, it's really interesting and amazing to see like Pat Gelsinger, the CEO of VMware up on stage with Andy Jassy. We did not expect to see him there. We actually got an interview with Sanjay Pudin, really like the number two guy at VMware and had a good discussion on theCUBE about this.

But right, first of all, I can take my VMware application and I'm going to be able to put it quote on AWS. Now of course, VMware on AWS doesn't launch until 2017 and the solution is kind of like, they don't call it cloudy even. Really it's, here's a special instance of Amazon that runs VMware but you now are in the same data centers, the same environments, I can throw a cable over the wall and be plugged into the rest of Amazon. So I now have access to all of those services and this is where we kind of look at it a lot of ways is Amazon has just stuck a straw in 500,000 VMware accounts and said, there's going to be this sucking sound pulling you towards the public cloud and you have access to all of these other services.

Oh, and if you want to migrate from VMware to just other virtualized environments without paying the VMware license, we've got some easy tools and services that allow you to do that. So we think it's a little bit dangerous for VMware but it does open up a big opportunity for VMware to kind of, you know, that whole cloud story that we've been wrestling with VMware on for the last bunch of years with the whole, you know, VCHS and vCloud Air and everything else in between.

So we won't get distracted about my feelings about vCloud Air but this is an interesting, I think, note for customers to take a look at how Amazon perceives hybrid or what their, I'll call it their winning strategy is. For customers who have bought into Amazon, this is definitely a win-win. From Snowball Edge to VMware Cloud on AWS, it makes sense, it makes the migration to cloud much more simpler, it's a enabling technology. But for customers who have a distinct view of what hybrid is, hybrid is a bunch of stuff that runs in my data center and a bunch of stuff that runs in AWS or multiple cloud providers.

AWS is creating, I think, some weight when it comes to adopting some of these services. If I adopt Snowball Edge, obviously I've committed to having this large data set in the cloud that I can move in but it's very hard for me to move out. And providing Edge Compute becomes simpler from an AWS consumption model but I'm kind of bought into this Lambda serverless stack that I'm just committed to. So it's one of those, it's lock-in but it's a type of lock-in that I think you have to go in with your eyes open to and be fully aware of the benefits and the pitfalls.

Yeah, absolutely, Keith. It's definitely something we're hearing from our community a lot is there's concern about lock-in. There are a bunch of people, including us at Wikibon, pointing out that we'd love to see Amazon embracing open-source much. They made a lot of hay around what they were doing with the MXNet, really for the kind of machine learning, artificial intelligence type space. And they did have a suite of open-source solutions they're working on around containers but did not inherently embrace Kubernetes, which is something that a lot of the community was hoping they'd do.

Oh, and they hired Adrian Cockroft who is one of the most well-known cloud and open-source proponents out there. So we expect to see more on him. But that being said, concern around lock-in. And I've been meeting with a lot of small companies that are trying to help customers manage their data, manage their operations in what the hybrid world, which is, it's my on-prem, it's public cloud, if not multiple public clouds. Oh, and don't forget all my SaaS. So how do I manage, where is the truth for the data?

Who has access to be able to see the data, change the data, make sure that I have consistency across all my tools? And Amazon's answer, like every other vendor, is like, well, the more you put on us, the easier that's going to be. But people are using so many different tools and services. While Amazon is a big winner, there's lots of players out there. And every vertical's got their own applications and their own needs that's going to cause me to have a highly fragmented solution.

And it definitely is not simple when I look at the marketplace. So this isn't a complicated, I think, market scenario compared to what we've been accustomed to in the past. All the way from the IBM mainframe days and then when IBM started to get competitors like Unisys in the market, there was kind of a stabilization around some level of standards. IBM PC to now, then there's obviously open source in Linux and Windows. And now even Windows and Linux are starting to play nice together.

NET platform across all platforms. The market eventually drives the standardization and interoperability. AWS, 14 times the size of its next, I think Jassy said, its next seven competitors combined? What's going on? I think it was actually that they're now bigger than their 14 next biggest competitors. I don't know that they have the orders of magnitude bigger that they used to, but whatever it is, they're the leader and they're way bigger than everybody else. Yeah, so what's going to drive, like when Adrian goes through and he proposes some type of, which I think he will, because back in his Netflix days, stuff like Netflix's Chaos Monkey open source project, that stuff has been immensely helpful to the overall cloud industry and customers as they adopt cloud.

What's going to be his rallying cry to help Amazon become more interoperable when it comes to their platform? I mean, I don't get the market driver for them to say all of a sudden, oh yeah, we'll make it easier for you to run serverless across us in Azure. Yeah, yeah, no, I mean, Keith, you bring up a great point. I mean, number one is, Amazon's really become the de facto API for a lot of environments, like S3 on the storage standpoint. There's lots of solutions that really just integrate deeply with that.

So do we integrate on the API? And that's kind of where they are, but it's open to an extent, but it's not really open source. But I think kind of containers is an area where they can definitely be more open and they've been involved in some of those environments. I'd love to see serverless get a little bit more open. IBM has OpenWhisk, which is an open source option out there, but Microsoft has their functions. Google has their cloud functions. Amazon's got Lambda out there.

And we saw this year how much Lambda is just spidering its way into lots of the solutions that AWS is putting out there, like the Greengrass that you mentioned. And customers are super excited about it. I heard the sessions were just overflowing. It really went from this seems really cool to wow, this is something we can deploy really easy. Oh, and by the way, Keith, they gave out 25,000 of the Amazon Echo Dots. So you've got Alexa on that. So I've got one at home.

My family's loving it. Capital One did the sponsorship of it. And at the show, they had little hackathons and stuff like that that people could create functions to be able to write to Alexa. So expect the skills to just grow on that. And I mean, that's like a really interesting kind of consumer edge use case of using this functions as a service or serverless technology that you can see how it uses. So God, you put it in the consumer's hand and that'll eventually bubble through to the enterprise really quick.

And I think you hit on a key point, which is this grassroots adoption for serverless. I think Lambda is so far ahead from a just, AWS is so big and Lambda is so intertwined with AWS services and AWS makes it so appealing to use that I think the industry is probably going to have to do what they're doing around S3, which is just make their serverless functions Lambda compatible, whether it's through a common API that Amazon publishes, which I doubt they would because why would they?

Or through just brute force re-engineering. But on the container piece, I think you make a really great point. The Kubernetes on its own has like every, I remember when Mesos was leading kind of that conversation and Kubernetes was kind of this weird science experiment from Google that everyone just said was too hard to now. If you Google container orchestration and container management, Kubernetes is the name. And I think AWS will hopefully get forced into a model where eventually they will have to adopt it because that's what customers are asking for.

All right, Keith, I think we need to put a call out to the community. How do we get Kelsey Hightower and Werner Vogels up on a stage, hugging it out and saying how Amazon and Kubernetes are going to work great together, right? That way we can make that happen. I think we can make that happen. I mean, Kelsey just tweeted, I retweeted him yesterday. He tweeted out that VMware containers, full-time containers is awesome. Integration with Kubernetes, he loved it.

I'm like, oh wow, if that's not a co-signing of a strategy, I don't know what is. So if he can embrace containers on VMware vSphere using Kubernetes, I'm sure him and AWS, Werner Vogels and the W, pronouncing the W as a V always throws me off. But those two together on stage, maybe we're at Interop or something, I can see it happening. Awesome. All right, so let's close this out. What are your big takeaways for customers when it comes to AWS reInvent?

What should customers really be paying attention to from the key announcements? Yeah, so much is going on. What Amazon is trying to make sure is that whatever a customer wants to do, they've got a service that either they offer or it's available in the marketplace. We didn't get to talk about the ecosystem much, but I really saw a huge growth of the system integrators that are there, the channel partners that are there, not to mention the thousands of software companies that are there.

So like kind of Intel had been the platform in the past, like we've seen the ecosystem rally around Microsoft and VMware, Amazon is really a platform that everybody's rallying to. It is moving from most customers from moving from kind of tactical, trying to figure out this cloud to strategic, understanding where it fits, understanding what I need to do from an IT standpoint, and I'm going to have lots of places I can turn to to make sure that I can take advantage of this solution without having to have everybody become experts.

And by the way, lots of resources being put in to all those training things. I saw online lots of mentions of like a cloud guru, what Amazon's doing to really help train people up, because there's an opportunity for the next generation of jobs here. All the people that have been sysadmins or architects, cloud is the opportunity, and if you've got the skill sets, there's people looking for it. Everybody at the show was hiring, so it's always exciting to see this stage of the ecosystem where it's growing gangbusters, there's lots of opportunity, and more people are getting jobs.

So we are still in the early days, there's still maturity we want to see, there's still kind of gaps, and Amazon is not the only solutions out there, so it's a great time to dig into more, learn about it, and go into 2017 with a decent plan. Well, Stu, I really appreciate you jumping on and having this conversation with me. Where can people stalk you online, and are you done traveling for the year? Is there any more conferences that the queue will be in?

So I think we are actually done with our travel schedule. I don't have any flights scheduled for the rest of 2016, so I'm super excited for that. Keith, I hope it's a long time before I have to do a red-eye in a while, like yourself. So yeah, easiest way, hit me up on Twitter. I'm just at Stu, as to you on Twitter. com is kind of the research website. tv for the video. We've already been posting a bunch of the shows we're going to be at in 2017.

Always love meeting new people in the community. Feel free to reach out to me with any questions, and always great to catch up with you, Keith. With that, Stu, thanks a lot. You can find the podcast on iTunes. Rate us, it helps with discoverability. You can find me on Twitter, at CTO Advisor, that's spelled the UK way with a O-O-R. Talk to you guys soon, thanks.