Last episode of Mobycast, we began our post-coverage analysis of AWS re:Invent 2019. With a major theme of “transformation”, we walked through some of the key advancements being made by AWS to drive innovation now and into the future. From supercomputing to networking to AI and ML, AWS is proof that there is “no compression algorithm for experience.”
In this episode of Mobycast, Jon and Chris conclude their special two-part mini-series on this year’s re:Invent conference. We finish recapping the big keynote sessions and highlight the major themes of this year’s show. We close it all out by sharing our most important takeways that you need to know.
In this episode, we cover the following topics:
- Recap and analysis of Andy Jassy’s keynote, including:
- The theme of this year’s keynote is transformation, presented via 6 theme songs.
- “The hunger keeps on growing” (Dave Matthews Band, “Too Much”)
- Storage performance is growing much faster than compute/memory (6x faster since 2012). This is enabling new innovations like AQUA for Redshift, making it 10x faster than any other cloud data warehouse.
- How the new Ultrawarm for Elasticsearch tier reduces cost by 90%.
- AWS helps again with undifferentiated heavy lifting by offering a managed Cassandra service.
- “I would walk 500 miles” (The Proclaimers, “I’m Gonna Be (500 Miles)”)
- AWS is one of the best places for AI/ML across all layers of the stack.
- SageMaker is a vibrant ML platform that is rapidly evolving into the next generation developer desktop.
- Can AWS really automate code reviews with its new Code Guru service?
- Using the power of AI to search for enterprise data with Kendra.
- “Break on through to the other side” (The Doors, “Break on Through”)
- If you can’t come to AWS, AWS is coming to you! AWS infrastructure is everywhere with VMware Cloud on AWS, AWS Outposts, AWS Local Zones and AWS Wavelength.
- Recap and analysis of Werner Vogel’s keynote, including:
- The problems and limitations of classic virtualization (Xen-based).
- How the Nitro System takes a microservices approach to computer system design.
- The importance of Firecracker as a next generation VM.
- A case-study of the Elastic Block Service (EBS) architecture and how AWS is using a new technology called Physalia to battle the CAP theorem.
- You can now learn more about how AWS designs and develops software with the new Amazon Builders’ Library.
- Key takeaways
- The Nitro System is a key technology that is enabling AWS to break through barriers. It will provide us with powerful new capabilities to tackle previously unsolvable problems.
- AWS is everywhere, extending all the way out to the edge with IoT/Greengrass, and everywhere in between with on-premise (AWS Outposts), near-premise (Local Zones) and even integrated on the 5G network with AWS Wavelength.
- AWS is a leader in AI/ML, with Sagemaker becoming a next generation developer platform.
- Analytics is a big priority, with focus on Redshift and data lakes.
- Quantum computing is no longer the stuff of science fiction. It will be here sooner than you think.
- AWS re:Invent
- AWS re:Invent 2019 Session Videos
- AWS Nitro System
- Elastic Fabric Adapter
- The Amazon Builders’ Library
- Amazon Braket
- Amazon Braket – Get Started with Quantum Computing
For a full transcription of this episode, please visit the episode webpage.
We’d love to hear from you! You can reach us at:
Stevie Rose: Last episode of Mobycast. We began our post coverage analysis of AWS re:Invent 2019 with a major theme of transformation. We walked through some of the key advancements being made by AWS to drive innovation now and into the future from super computing to networking to AI and ML. AWS is proof that there is no compression algorithm for experience. In this episode of Mobycast, John and Chris conclude their special two-part mini series on this year’s re-invent conference. We finish recapping the big keynote sessions and highlight the major themes of this year’s show. We closed it all out by sharing our most important takeaways that you need to know.
Welcome to Mobycast a show about the techniques and technologies used by the best cloud native software teams. Each week, your hosts, John Christensen and Chris Hickman pick a software concept and dive deep to figure it out.
Jon Christensen: Yeah, personal story. This is all really interesting to me because a friend of mine who you met Nate Siemens is an architect over at NASDAQ and he’s been involved with NASDAQ, which just has billions and billions and trillions of pieces of data that they look at and try to figure out because he’s in the risk analysis area of NASDAQ I think. So he’s got to look across transactions and think about things. I guess Rob Hunt, who’s the VP of NASDAQ, did a session, I didn’t know he was doing this, but he did a session at re:Invent this year and he talked about their journey from on prem. And that was when Nate was doing Cassandra stuff. How [inaudible 00:01:44] to a cloud native data warehouse using Redshift. And then they went off of Redshift to using historical data in parquet in S3 with Presto and spark.
And then they went back to a Redshift cluster with historical data and X3 accessible via spectrum. So AWS lost NASDAQ’s business and they all… I mean it was still an X3, but lost probably a lot of money, like that Redshift cluster that they turned away from was probably making them a good amount of money that they lost when they moved over to S3 with Presto and spark in and they won that business back when they brought in the spectrum capabilities. So super interesting.
Chris Hickman: Yeah, so much change going on there. And just part of this definitely goes back to Amazon’s one of their core principles of just customer obsession and just really listening to what customers are saying. Right?
Jon Christensen: Yeah. And I’m sure NASDAQ can move the needle with Amazon.
Chris Hickman: Yeah, absolutely. Yeah. So another thing as you pointed out is just storage performance is growing much, much faster than compute and memory. So since 2012 there’s been 12X improvement in storage versus 2X in CPU and memory. I mean, we talked about this before, just MacBook pros where you can get an eight terabyte SSD inside a laptop computer, it’s more pounds, right? Just crazy and mind boggling. So, another new announcement that they have was an Aqua for Redshift. And so this is, basically they say it allows Redshift to run 10 X faster than any other cloud data warehouse. And so they built a high speed cash on top of S3 and it’s using this custom AWS design chip to do so. If you’re into using Redshift and you’ve got lots of data, like this is obviously something that’s going to be very, very useful.
Jon Christensen: Cool.
Chris Hickman: Ran in and out. A couple more announcements in the analytics space, they have ultra one for elastic search, which is instead of having you manage like storage tiers and whatnot, that’s doing it for you, where it’s key, you can have your index as be the hot data and then other data be more in this warm tier. And they just saying it’s going to save you 90% and you’re just going to reduce costs by 90%. So if you’re using Alaska search sounds pretty good. Right?
And then they announced they now have a managed Cassandra service. So for just like last year they did the managed service for Kafka this year they’ve now got a managed Cassandra. So for folks using that, this takes away that undifferentiated heavy lifting and let AWS do it for you. All right. So that was theme aspect four. So the fifth one was I would walk 500 miles, which is a song from the proclaimers. I’m an oldie but goodie. And this one was really all about like what they’re doing in the AI and ML space. And again, no surprise, like this is a focus, I think.
Jon Christensen: Yeah, I mean for me the main thing here is like Sage maker is here to stay and it’s pretty, it’s good. It’s a good way to do your experimentation and prepare your models.
Chris Hickman: Yeah, absolutely. So, I mean, just by sheer count, there were six new features or services all around Sage maker that he announced during this talk. And then some of the other AI services, they announced a fraud detector, which is a real time fraud detection system code guru, which they pitched as like this is-
Jon Christensen: [crosstalk 00:05:22] everybody’s so excited about it. But I don’t think it’s very real. I don’t think it is. It costs a lot of us. I heard a lot of hot hallway talking about code guru.
Chris Hickman: Yeah, sure. I mean like Pete who likes doing, I mean I actually enjoy doing code reviews, but I don’t think… most normal people wouldn’t… That’s not so much fun. Right? So LGTM looks good to me as their comment and PR approved. And so the idea of having like someone else… A computer do the code reviews is definitely a pretty attractive. But we’ll see how this shakes out. I think right now this is really for the most part, static code analysis along with a profiler. But I’m sure they’ll continue to work on this and improvement and who knows what it will be a couple of years down the road. A couple other things. Contact lens for Amazon connect, which basic… So Amazon connect is their call center product software and contact lens is this now ability you can auto transcribe and analyze customer calls.
So pretty, pretty cool. Pretty impressive. So you’re handling calls from customer service support and now contact lens will automatically transcribe those calls and be able to analyze them, whether it be for sentiment analysis or just finding querying on that data or whatnot. It’s all you get it just by checking a box now. So pretty cool. And then Kendra, which we talked a little bit about before where this is now a search across all your enterprise data documents, FAQs, whatnot and then using natural language to query it, right? And to find it. So it’s all being backed by this AI.
Jon Christensen: I was so excited about Kendra that I went and tried to play with that a little bit and be, now my takeaway is that it’s got a long ways to go to be useful. I couldn’t even really connect it up to anything because right now it only connects to databases and something else, which I didn’t have anything ready to go. Right. Like I didn’t have a production database where I was like, Oh, I’m going to plug in Kendra, I’m just going to go ahead and do this. So yeah, it feels like there’s a lot of room for them to add. And actually there are several things that are on the roadmap for Q1 2020 that looks like it’ll become more interesting.
Chris Hickman: Yeah. Yeah. So yeah, I mean just from a vision standpoint, like-
Jon Christensen: It’s cool.
Chris Hickman: This is what they’re driving towards. And-
Jon Christensen: Love the vision.
Chris Hickman: Yeah. And they’ll get there. And then the last section of the keynote was-
Jon Christensen: Quick point on that. Yeah. Google tried this, they didn’t succeed at it. They were trying to do it with an appliance from what I recall, like put this appliance in your data center and they just didn’t understand how enterprises work. And the anecdote I heard was that they were surfacing data to people who shouldn’t really have access to that data. So customers were unhappy and they couldn’t really figure out how to partition the data properly in terms of our back rules and visibility. So I think that… AWS is much closer to probably being able to understand customers and understand the enterprise.
So I think that they’re going to have a better shot at it if Microsoft is also doing this. I think they’re probably going to have an advantage over AWS in terms of just understanding how enterprises work, how they want their data to be searchable and who should be to see it. And also how it’s hidden away and what’s available, what’s not.
Chris Hickman: Yeah, indeed.
Jon Christensen: Cool.
Chris Hickman: Absolutely. So, and then the sixth part of the keynote was based on the door song break on through. So break on through the other side, which at the end of that it was like, “Hey AWS, we’re just going to bring the cloud closer to you to where you’re at.” And so Andy ran through like the various options that they have now. So there’s the VMware cloud on AWS, right? Which is where you can actually run VMware on AWS. So most folks on prem, they’re using VMware to run virtual machines. And so it becomes as like this really seamless migration path of getting onto the cloud by being able just to run the same VMware VMs inside AWS. So that’s one.
Another one is outposts. So they announced outpost last year at re:Invent. This year there was a lot of talk about it because it’s now generally available. So this is where you can get a rack of AWS hardware that is delivered to you, put into your on prem data center and it’s installed, managed, monitored all by AWS. You can access it through the normal AWS console. All made possible based because of the nitrous system.
Jon Christensen: I’m getting one for my office.
Chris Hickman: Sure, why not? Yeah. Got a few hundred thousand dollars sitting around got to spend it, taxes and eight year purchases to offset gains. Yeah, so there’s output. And then another new announced with is for something they’re calling local zones and local zones. Basically what it is, is it’s outpost running in an Amazon managed building.
Like this seven 11 of AWS.
Jon Christensen: Sure, sure.
Chris Hickman: But it’s like [inaudible 00:10:49] like have you ever seen like an Amazon pickup location? They have [crosstalk 00:10:55]
Jon Christensen: A whole foods Mark.
Chris Hickman: Yeah, yeah, exactly. [crosstalk 00:10:57]
Jon Christensen: You mean the lockers?
Chris Hickman: Yeah, yeah, probably. I mean not that, but it’s the same idea. Like let’s bring the cloud into your local quickie Mart.
Jon Christensen: Yeah. So with outpost you have a reduced surface area of services, right? That are available. So instead of like 175 it’s like more on the order of like 15 of the core services. But yeah, it’s extending the region to be even closer to you. So like the first local zone they’re implementing is in Los Angeles. So for folks on the West coast, instead of going all the way up to Oregon or US West one I think is what San Francisco area.
If you’re down in that area, like going to the LA area is going to be faster and less latency. So there’s the advantage there. And so LA is the first one. There’s going to be more. And it just pops, basically pops that are just closer to you by the various services. And then the last one that they announced is this new offering called the wavelength and wavelength is basically marrying up AWS hardware pro services with the 5G providers in the network. So this is really aimed at the mobile devices and then other devices that are connected over 5G where they’re really addressing the latency because right now, I mean there’s that latency of like if you are on the 5G or you are on a cellular network, you first are going through all the solar infrastructure, right?
And going to the… I think there’s like a city hub, then there’s a regional hub before it actually gets onto the backbone of the internet. Right? So there’s all this latency before it actually finally makes it to the cloud. And so they’re saying… Well they’re addressing that by basically by taking an outpost I think and putting it right there at the cellular providers networking infrastructure so that you don’t have to go out over the backbone. And so you can reduce latency quite a bit for these applications that may really benefit from that. So whether it be like gaming or something else, this is going to be, it could be very, very interesting.
Chris Hickman: Right, right.
Jon Christensen: They didn’t make a big… During this part of it, they brought on stage, the CEO of Verizon wireless to really just talk about 5G and it’s going to be interesting because I think I haven’t been too terribly excited about 5G because it’s like, we have 4G and like you can walk down any street, stream video like no problem whatsoever. It’s like it feels like we have all the bandwidth we need.
Chris Hickman: Right. I hear you. One of the things I think that 5G could help with immediately it would be self driving cars. Like if you could start to put some processing power and to stay like all the stop signals that like every interchange then you can start to have safe zones for self driving cars. Like you know that you’re on the road, you know like what the signal is saying. Like you know that you’re not driving up onto a sidewalk or something because you’ve got like processable information nearby that you know has low latency and is is correct and accurate. So I think that’s an area that could see some of that innovation.
Jon Christensen: Yeah. And at the end of the day and they just be like this concept of like you’re just… This is the new network, right? It’s no longer [crosstalk 00:14:35] you’re not going over fiber, you’re not going over copper. It’s just like everything can be wireless over the cellular networks just because it’s got the bandwidth and it’s fast enough and the latency is low enough where it just works and no wires is better than wires. Right?
Chris Hickman: Right. So we’ve gotten long already, but I think maybe we just blow this one out a little bit. Chris might as well finish up talking about re-invent and there’s a little bit more to discuss I think. What do you say?
Jon Christensen: Sure. Yeah. Yeah. So why don’t we move on to the last keynote we wanted to talk about which is Verner vocals keynote. For me this is a little interesting just because it was shorter this year and also it just… In previous years, like burner, there was a lot of announcements, especially in the core technology developer ecosystem space. Like last year there was a whole bunch of stuff around Lambda with Lambda layers and the custom runtime and there was just a bunch of other stuff in that space, right? And this year in his keynote, there wasn’t any announcements really other than the Amazon builders library. And so I mean it’s cool, right? I mean it’s a collection of documents that basically are like giving you the inside knowledge of just how Amazon builds software, right?
So if you want to know about shuffle sharding and how that leads to better availability, right? Like there’s now a paper in that and in the Amazon builder’s library that you can go read about, but it feels like, I mean, a lot of that information I think was already out there and one form or another. And it’s just like, now here’s just another place. So we have like, we have the well architected tool to go to now we have the quick starts area that they have there is, I mean there’s several different places now where there’s these repositories of knowhow and templates and quick starts and getting started and whatnot. And this is just adding to that, this is just feels like it’s probably a more of a-
Chris Hickman: It’s like organizing.
Jon Christensen: Yeah. And like the information that’s here is going to be more like from just an architecture maybe in design standpoint and less of a just here’s how you would put these pieces together, right?
Type thing to roll this out. So that was really the only announcement here as part of his talk, the talk really boil down to a couple different sections. One was on just virtualization and where they’ve been and where they’re going to. And that ended up just pointing out like, “Hey, we started off using Zen virtualization and we talked a bit about this in the VMs versus containers, mini series on Mobycast where how VMs are implemented. So it talks about like using Zen as the original hypervisor and the problems with that and how they’ve now have this nitros system, which is just so much better on so many different levels. And I think we’re just going to end up having to, maybe we’ll talk about that on a future episode of Mobycast.
Chris Hickman: [Crosstalk 00:17:41] this one.
Jon Christensen: I know I mean there is actually… It is really impressive. There’s a lot of pieces to it and it’s probably, it’s just not something that we can talk about in three minutes, but just other than that, like this is the future. It’s enabling lots of innovation and all the new instance types are using it. We’ve talked about the benefits that we get from a networking standpoint. There’s a lot of benefits that you get from a security standpoint and other performance standpoint. So really, really, really key a crown jewel for AWS.
Chris Hickman: For me, I think it’s worth thing at least one thing at least in terms of how I think of it. It’s like all this stuff that’s virtualized, like your desk, like your networking, like your IO, like your memory, all that stuff that gets virtualized like that used to be managed by this hypervisor that lived, that ran in your processor and now it’s like, Oh what if he could take that and put it somewhere else and just have a whole process or just for dealing with that stuff. And that’s essentially what nitro is.
Jon Christensen: Yep.
Chris Hickman: Cool. We didn’t deep dive on nitro.
Jon Christensen: Yeah. Not quiet but yes, hopefully we’ve laid the groundwork for that. And then along with nitro I just talked about firecracker and again highlighting that is like again, one of the crown jewels that they’ve been working on that’s really starting to pay off. So we talked about that. This is truly a lightweight VM, so it combines the best of both worlds. It’s the security and the isolation of the VM, but the performance, the speed of a container and the Lambda service is currently running on firecracker. I already thought that far gate was using firecracker. I thought that’s what happened when they had the big price cut-
Chris Hickman: Yeah me too.
Jon Christensen: [inaudible 00:19:31] end of the year but it turns out no, they’re not. They’re not on firecracker yet. They’re working on it. It’s coming soon, but as of yet far gate, it is not on firecracker. They’re still running there. They’re spinning up a VM for every container on far gate right now, which is not efficient at all. It’s very expensive. Right? So once firecracker comes along, they’re going to get much better density and prices will come down quite a bit.
Chris Hickman: And that’s when we’ll see the cloud run competitor that we talked about before, the Google cloud run thing that can automatically scale up and down an image, a Docker image. Another thing though that that stuck with me from that firecracker announcement or just from discussing firecracker was invent a way to think of it is that there’s all this crust in a VM, all this stuff that computers have to do that you just don’t need for most computer, most applications. So just things around dealing with lots of different devices, things around managing tons of different users. If you’re just running a single process, like you don’t need all that stuff. So that’s why it really is. There’s just less the virtualize and that’s why it really is a lightweight virtual machine. They’ve literally just stripped out a bunch of stuff from a typical operating system that most processes don’t need.
Jon Christensen: Yep. Yep. Yeah. It’s like you don’t need to virtualize the display. There is no display. So that’s just not part of it. Yep. So it’s very much purpose-built and we reap the gains from a performance simplicity standpoint.
Chris Hickman: We cover a lot of information here on Mobycast and if you’ve ever wanted to go back and remind yourself of something we talked about in a previous episode, it can be hard to search through our website and transcripts to find exactly what you’re looking for. Well now it’s a lot easier. All you have to do is go to Mobycast.FM/show-notes and sign up. We’ll send you our weekly super detailed outline that we use to actually record the show. And a lot of times this outline contains more information than we get to during our hour on the air, the signup and get weekly Mobicash cheat sheets to all of our episodes delivered right to your inbox.
Jon Christensen: And so then the rest of his talk was really an architecture case study for elastic blocks storage. So EBS and so just talked about how EBS has a cell based architecture. I think we’ve talked about this in the past. This is a very core to how AWS build services is they take the cell based approach where you’re partitioning up your architecture into individual groups that are fully contained, that stand alone by themselves, right? And then you use things like partitioning and sharding to break it up into cells. But the idea is that you’re reducing your blast radius so that if a database dies, then it only affects that cell. It doesn’t affect the entire service. And so EBS is uses this cell based architecture, but they did run into the problem where they have to keep state of where the data is and they call that their config master.
And that ended up being a single point of failure, right? So it’s like how do we deal with this? And so this is the first time that they’ve ever talked about a technology that they call FA Zalea.
Chris Hickman: FA Zalea.
Jon Christensen: FA Zalea. This is their approach to solving that problem. So really what it is, is it’s breaking up that database into millions of tiny little databases and they think of this as colonies of cells or like the whole-
Chris Hickman: Jellyfish model.
Jon Christensen: Yeah. What’s interesting about this is that it really, like this design came about from their thinking about the cap theorem. And so the cap theorem says, so there’s consistency, availability and network partitions and you basically can have two out of those three things. So if you want to be… You can’t have all three at the same time, right?
Is really what the cap theorem is telling you. So they knew that for EBS consistency was very important. Right? Because it would be bad if you’re like reading from the disc and sometimes it’s there and sometimes it’s not there. And then availability obviously very important as well. So in order for this to work, they knew that they had to reduce the probability of network partitions, right? So they had a simplify, the networking of this and reduce basically the size of the network between the nodes in this cellar colony. Right? So to do that, so to reduce the chances of these network partitions, that’s where they went this idea like what works can millions of these really small databases and we’re going to optimize the placement of those database nodes to be as close as possible to those nodes that need to read and write to that database. Right?
So you’re reducing the size of the network between them. So an interesting approach, it was again, like an architecture level type talk about just a really interesting challenging problem and how Amazon is going about and solving it. And so they’ve done this and they’ve implemented and they’ve seen their EBS error rate go down significantly and they have much better consistency now with those error rates because they’ve gone from this more centralized model to these millions of little tiny databases and reduced their profitability of network partitions happening.
Chris Hickman: So what was your takeaway from that part of the keynote, Chris?
Jon Christensen: From the talk about this-
Chris Hickman: FA Zalea.
Jon Christensen: FA Zalea?
Chris Hickman: Mh-mm( affirmative.).
Jon Christensen: I think it was just like just demonstrating AWS is just, again, there’s no compression algorithm for experience and something like this really can only come from going through all the trials and tribulations and, and getting to the scale that they’re at that is forcing them to address these problems and come up with really interesting solutions, noble solutions to things. Right? And so it’s just showcasing that breadth and that depth of experience that AWS has.
Chris Hickman: Yeah, I guess my take away is a little bit more cynical. I mean I agree with you, that’s true. But my take away is a little bit more cynical, maybe genetic, cynical illness. It’s like, okay, so at the beginning of the keynote, Michelle Verner in a video and he’s trying on t-shirts and they’re talking about how, whatever tee shirt he wears drives virality of some tweets or like pushes up like tweet hashtags, right? Like people are talking about it. So that when I got from that is like he’s a popular celebrity. He’s a celebrity among developers. He absolutely is. He’s got TV shows, right? This guy is a celebrity within a reason, within a certain group, a niche celebrity. So then you get them on stage and you have him talk really deep down technical telling you some secrets and you haven’t go just over the heads of almost the entire audience.
Just a little bit like it’s hard to understand him because of his accent and every once in a while you can pick up something like oof FA Zalea, yes, that’s very cool name and Oh yeah, blast radius. I get that part. But like he gets into the cap theorem stuff and it’s like just went over the heads of everybody and like but I get you again when he’s talking about bringing the little databases close to the actual data. Like everyone understands that part. And then the feeling that everybody’s left with this man, that guy’s so smart. And that’s the point. That’s it. That was to make developers frothy. That’s all it was. It wasn’t anything else. It was just marketing. Just put a guy that’s famous up in front of people and talk about fancy schmancy stuff so that they get all frothed up and go home and say how awesome AWS is. That totally is it. I mean that’s maybe the point of the whole conference, but it’s not right. Like we have actual jobs to do and we have to get our work done and we have to learn how to do AWS stuff. But the cynical me, like that was the takeaway from that keynote is like there wasn’t anything new, there wasn’t anything different that he talked about. So let’s just show off my smarts so that you love me more. And you tell your bosses about me.
Jon Christensen: Yeah, definitely a little bit more cynical view than mine. I think paradoxically they’re both viewpoints are saying the same thing. Just coming from different angles. Right. So it’s like, Hey AWS they’re faced with some just problems that most of us just, well we’ll never see. Right? Because we’re just not working at that scale.
Chris Hickman: And so the actual detail there, it doesn’t really matter to us that much. Right? But it’s cool. It’s fun.
Jon Christensen: Yeah. I mean like I said, for most people it’s going to go over their heads and they’re not. I really appreciate it other than it is a cool name and like yeah you’ve convinced me. That seems really, really hard and I don’t understand it. So most of the people in the room probably fit in that category, but-
Chris Hickman: Not most of our listeners though, right? Because we’re doing it, we’re bringing everybody up. Yes. [crosstalk 00:29:07]
Jon Christensen: Hopefully we are bringing you along. So all right. So that covers the keynote, the three major keynote, we’ve gone through all that stuff. So maybe now we can just tease out like, okay, just what’s the takeaways from this and what are some of that? Maybe the eye-openers. And so for me, nitro system-
Chris Hickman: Nitro, nitro.
Jon Christensen: We’ve talked a lot about that in the recap. And again, it’s a crown jewel for AWS. They’ve invested a lot into it and it’s really going to be like the backbone for innovation going forward and it’s really going to enable a lot of amazing capabilities, in instance, types, new applications, just immense amounts of computing power and then really just taking it AWS just all over, just everywhere it is, which leads into another like takeaway from me was just AWS is everywhere now.
So it’s all the way out on the very edge with things like Lambda edge and their IOT initiatives. Like with Greengrass, it’s on prem now with AWS outpost, it’s near prem with AWS local zones, it’s in 5G now with wavelength. And then of course it’s still in the cloud, right? With what we’re used to. So it’s literally being… it’s coming to you, right? So if you’re not going to come to it in the cloud, it’s going to come out to where you are and be closer to you. And so it’s just this really broad spectrum of coverage and that will just continue. Right? It will just continue to blend in. I mean I wouldn’t be surprised if they come out with AWS laptops, right? That have [inaudible 00:31:04] right into it right?
I mean why not? Right. And it’s like the perfect developer machine if you’re working on AWS or something like that. Another big, big, big theme for me, which is just like the importance of AI and ML and particularly with Sage maker. So like I said, they had a lot of announcements around Sage maker and at the end of the day they’ve turned Sage maker into a complete full up developer platform. So they now have a Sage maker IDE. They have-
Chris Hickman: Debugger.
Jon Christensen: Debugger. You can save your projects as a collection of experiments. You can collate all the artifacts around those experiments. There’s operational support for like monitoring your models once they are deployed to make sure that they’re not drifting.
Chris Hickman: I mean this is huge Chris. Think about this, as you said to me, AIML practitioner capability, like being able to get in there and do stuff with it, not necessarily coming up with new ways of making models but training models and according to like some recipe books and things like that is baseline expectation for developers now. Right? You got some data, better person, a AIML to get insights out of that data to some categorization or some prediction and AWS like they have cloud nine and people are definitely using cloud nine a little bit, but they’re really not touching code vs code or visual studio in terms of like IDE love, but they really have a shot at being the IDE of AI ML. They really do. And that’s maybe more right now, isn’t it?
Jon Christensen: Yeah. So I mean, I think up till now, like developers, people in this industry understanding AIML being able to use it as a practitioner has been optional. Definitely going forward in the future. It is required right there. This is changing so rapidly. It’s becoming so highly capable and advanced and it’s just going to be ingrained in everything we do. Like you have to be up to date on this, you have to use it, you have to be able to weave it into your work. Otherwise it’s just going to leave you behind. So AI ML, like get on it people. If you haven’t to like go do it. So, and there’s a lot of great ways to get into this, right? It’s a huge field and there’s lots to learn, but you don’t have to do it all at once.
You don’t have to go get a college degree in this, you don’t have to go do full up courses or whatnot. I mean, just start off like taken off bits and pieces. So go play around with deep racer. You don’t have to go buy the $400 car, right? You can do it all via the simulated environments, but go do the deep racers-
Chris Hickman: Was there about to be an admission of [inaudible 00:33:48]? I mean you just dropped off.
Jon Christensen: If you remember last year during our coverage of re-invent, I mentioned this like really like you guys are making this big announcement. Like it’s a big deal at deep racer, it’s like a $400 and then car, toy type thing. Like who’s going to go spend $400 a pair on the toy and you’re going to have like these competitions and it’s like, okay, whatever. It seems out of whack with everything else.
And so now I’m taking a little bit of a pivot there where it’s like, you know what? This is probably a pretty easy way. Like, don’t go by them. You don’t have to go buy the car, but go use the software and tools, the simulations to go learn. This is an opportunity for you to go learn about how reinforcement learning works, right? So go play around with it. And by doing so and by doing a simulation of getting a car to try to follow a track or whatnot, like you’re going to learn about that. And so same thing goes with deep composer, right? So deep composer is a new thing they just launched. It’s a keyboard where I’m using AI to basically write music, right? And to play it. But you can also do that without the actual piece of hardware and it’s teaching about Gans. Right?
So general adversarial networks I think. Right? So another part of AIML that you can go learn about, just these tools and it’s going to be fun, right? And same thing goes so deep lens with something they announced two years ago, which is teaching you a bit about like how some of the AI ML techniques work for image recognition, right? So go, start doing stuff, start playing with this stuff, right? To come up to speed for it. Go see it. I don’t know if Coursera still has the course on introduction to machine learning and going all into just the basics of it on how it works with… That was a really good course that was free. Again, I don’t know if it still is, but if it is, I mean I would definitely go give that a look. It’s 15 hours of work. I think you did that one, didn’t you?
Chris Hickman: Yeah, I did. Yeah.
Jon Christensen: Yeah. So start… if you haven’t looked into AI and ML and if all this stuff is new to you, you jump on it because I don’t know when it is, if it’s two years from now or three years from now or five years from now or even tomorrow like it is going to be pretty important to your job.
Chris Hickman: Agreed. Agreed [crosstalk 00:36:26] Oh go ahead.
Jon Christensen: I was just saying, so just wrapping up just a few more things. So analytics was just a big focus of this re-invent with, I was really surprised by just how much effort they’ve put into Redshift. I mean that’s been one of those ones that in the past doesn’t seem to have gotten a lot of attention. And this year it was, the spotlight was on it-
Chris Hickman: Yeah, it was.
Jon Christensen: Big time. So Redshift and data lakes, big, big, big focus for AWS. And then the last thing that really I’ve come around on is quantum computing.
Chris Hickman: Yeah, we did have to touch on this. This is so cool. I’m so glad that you came around on it. I mean, let’s tell this story. There was a guy from D-Wave who came to give a talk at Glu con and I think I might’ve even talked about it on the show, how it was so interesting to see that they were building a little IDE, a little software development kit for developers to be able to get some time on D-Wave. And then there was like a D-wave cloud. And I think at the time you were just like, yeah, come back to me when you can do something real with it. And now, yeah. What’s your thought now?
Jon Christensen: Yeah. Yeah. So I mean, I do remember that conversation and you were like, “Hey, isn’t this cool and aren’t you interested in?” And I’m like, “No, not really.” This is still in the labs at science fiction. Like this is not going to be real for tens of years, 10 years type thing. So it’s like, I’m not even like, it’s just not even on my radar. Right? There’s been so many technologies like that in the past where it was like, there’s lots of buzz around it and one of his like photon computing, right. Using light instead of electricity as the medium inside of instead of a circuit board and light. The advantage being that photons don’t interact with each other like electrons do. And nothing came of that.
Right. There was a lot of buzz around it. And like there was stuff that was built in the lab that seemed like it worked and whatnot. But it didn’t go anywhere. And actually with quantum, there’s been a lot of stuff there too, right? Like we see the articles about like, Oh quantum entanglement and like time travel and how they got like these two quantum particles to what they’re calling time travel, where one was able to communicate information to another one that was like 30 miles away. But it’s all under like these really weird lab conditions type of thing. So it just felt really science fictiony and at this re-invent AWS did announce bracket, which is their new service for quantum computing. So giving you access in the cloud to quantum computers like from D-Wave and Island Q and Righetti.
And so that got me looking at that a little bit more. And as I looked at it a little bit deeper, it was like, I was really surprised by how far along it really is. So started off with quantum computers, didn’t have much capabilities. They only had a few cubits. And so cubits are the like the basic building block for a quantum computer as opposed to bits. Right? So think of it, if you had to like a four bit computer, like that’s not very impressive. Right? It’s not going to do terribly much. And so it started off with just had a handful of these cubits. Well now the current state is that they’re up to about 50 to 70 cubits in a quantum computer and around a hundred cubits is when you can start using algorithms like the traveling salesman and solve those kinds of problems with a quantum computer.
Chris Hickman: Yeah, that’s big.
Jon Christensen: I mean that’s right around the corner. Right? Look at the way technology the way that we’ve seen it progress and it’s not necessarily linear. It’s usually exponential. So a hundred cubits is not that far away. At 300 cubits that’s where you can represent more numbers at once than the total number of molecules in the entire universe.
Chris Hickman: [crosstalk 00:40:40] what we’re going to do to my body?
Jon Christensen: Yeah, exactly right? So there’s so much effort being put into this now and I mean just so much resources and experts. I mean Google they’re at the forefront and this Microsoft is at the forefront and this and now Amazon is getting into the game as well. So there’s progress being made there. It’s no longer just in a lab, like if you want to go play around with quantum computing now you can do it and it’s-
Chris Hickman: Literally, I think it’s like D-Wave will give you time for free. Like you can actually go read up a little bit and go make something happen on the quantum like a real quantum computer. It’s pretty cool.
Jon Christensen: I think of it… I view this as like the same thing that happened with AI and ML like five years ago, it wasn’t too terribly practical and now it’s like we have like there was a freight truck that just drove itself from California to the East coast with 50,000 pounds of butter in it. Right? And it drove the entire way, the self driving vehicle the entire way itself. And it’s like if someone told you that was possible like five years ago, like that was going to happen. It’s like, huh? Really? I don’t know. I don’t know if I’d bet on that. And I think it’s the same thing that’s happening with quantum. I think where we’re at now, five years from now, we’re going to look back and say, “Oh my goodness, like this is pretty, pretty game changing.”
I think quantum is only good for certain classes of problems. It’s not going to be a replacement for digital computers. So it’s not one of those things where it’s just like, if you don’t learn how to be a quantum programmer, it’s just not going to work out for you. But you should like take a look into it and it’s something to be aware of and to know and it’s good to know like what does it potentially good for versus what it’s not and just as it’s going to be a compliment to the technologies that we have.
Chris Hickman: Right. Yeah. And because the pace of innovation is accelerating and we’ve already decided that in software it looks like the innovation is largely happening in the big companies in the public clouds. But as something I think some of our listeners will recognize when I say this is like, innovation is accelerating. So if you look to the past to gauge the pace of innovation and project that into the future and you will be incorrect, like you will underestimate how much innovation will happen in the next five years. If you look at the last five years and if you look at the last five years it’s like Whoa, so much happen. So yeah, more than that is going to happen in the next five years, which is just so hard to get your head around.
Jon Christensen: It is and then I will not even make any predictions about what is going to look like 20 years from now. Right. Because I really do think it’s going to be so completely different than what we can imagine. It’s-
Chris Hickman: So great. I mean we do and I think hopefully what’s coming out a little bit is how we were both of us inspired by re:Invent. That was great conference to be at and glad you could listen to us go on and on about it.
Jon Christensen: Yeah, absolutely. Yeah. Sorry it was so long this week, but lot to cover and really good linen. If you haven’t had a chance to get re-invent, definitely get it on your calendar.
Chris Hickman: Yeah, we’ll see you there next year.
Jon Christensen: All right. Thanks John.
Chris Hickman: All right, thanks Chris. Bye.
Jon Christensen: Bye.
Stevie Rose: (music)Nobody listens to podcasts outros. Are you still here? Oh, that’s right. It’s the outro song. Come talk to us at Mobycast.fm or on Reddit at r/mobycast.(music)