The #1 request from CenturyLink Labs readers is to hear about real-life case studies about Docker in production. This week, we are honored to have a fantastic use-case example of Docker in the real world.


Matt Butcher, Lead Cloud Engineer at Revolv

Matt Butcher is currently head of Cloud Services at Revolv... a crazy-cool home automation hub (internet of things) startup. Think Nest for everything else in the house. Works with your exiting devices (Belkin, Hue, Honeywell, Sonos, etc). He has authored of seven books and numerous articles on technology, and teaches at Loyola University Chicago.

How do you use Docker at Revolve?

We are still running many core services on Virtual Machines. We have played with a half-dozen Docker technologies and haven't yet committed to any one just yet. But we have replaced our entire CI/CD solution with Drone (a Docker based on-prem open-source CI/CD solution). It took about a week and a half. We had been using Jenkins and it was a nightmare. We are actively looking for more ways to incorporate Docker into production. We are seriously looking into using Amazon's Elastic Beanstalk with Docker, but haven't made commitments on it yet.

You wrote “Why Containers Won’t Beat VMs” a year ago. Do you still think that way?

At the time of writing that article, Docker was just 3 months old and not well understood and Virtual Machines were gang-busters. Containers looked like a faddy kind of toy. But I did not foresee the cool things that came out of the Docker community like CoreOS and Deis and the other micro-PaaSes. Containers are becoming a very elegant and compelling model for building applications. From a DevOps perspective, it is starting to turn out to look like Docker Containers are the right way of doing things.

What do you not like about Docker?

A week ago, it would have been the perpetual putting off of the 1.0. But now that is out. My biggest concern right now is that the tools around Docker are immature, but this problem is being solved by the community right now.

What is the biggest problem in real-life Docker adoption today?

The biggest thing is that right now if I want to deploy Docker, I still have to use Virtual Machines and then put Docker on them. It would be great to have pure Docker hosting from one of the larger hosting providers out there.

As a Docker user, are you interested/excited about the libswarm or libchan?

I am most excited about libcontainer. Seeing libchan which gives go channels at the network level is very exciting too. I am still not sure what to make of libswarm. It appears to be something more for the ecosystem than for end-users.

Are you using any orchestration or PaaS with Docker? Like CoreOS? Deis? Dokku?

First started playing with Dokku PaaS a year ago and I like the idea of minimalist build-your-own PaaS. I think it is very promising, but still takes hour to setup all the dependencies. We check into these projects every 2-3 months to see how it looks. So far it is not robust and mature enough, but we think it will be within 2-3 months from now. However we have backed off from PaaS and are going a little lower on the stack, closer to CoreOS.

You have blogged about using Drone for CI/CD… how has your experience been with using CI/CD with Docker?

Drone works by pulling stuff out of your git repository, build a custom Docker image with whatever dependencies you need (binaries and other), and then execute any arbitrary command you want. In Jenkins, even if you could wire up the code just the way you needed it, you were still running on the slave's OS which may or may not match up with production. From the moment the Drone container finishes building, we know that the production environment will match exactly the same state as dev/test. With Drone you can also spin up database containers that match production database containers. This creates a much more robust workflow for testing things than what has been available before.

How did you get into go? What do you like about go? What do you not like about go?

I started out doing Java for 10 years. Then I did PHP/Drupal for a while. When I joined Revolv, I joined as Java. However recently it felt like Java was nesting library upon library. With go I was impressed that I was able to build a remarkably robust application in go with just the core libraries. On the other hand, the fact that go compiles to a small size with low memory meant that I could use dramatically fewer resources.

In go, not everything may be easy, but everything in the language should be in the language. That is the suite spot that I wanted in a language. PHP had too much built-in and Java had too little, requiring you to use too many nested libraries. Go was a great middle ground.

Full Transcript

Lucas Carlson: Hello, and welcome. This podcast is brought to you by CenturyLink and CenturyLink Cloud.

Today, I am very excited to be interviewing Matt Butcher about how he uses Docker. This is a real world use case of somebody using Docker to get stuff done. I am very, very excited about it, and Matt is such a great guy.

He is a senior software developer and an author. He has written a lot of books. I've written a couple books so I know how hard it is. It's amazing, this guy is a machine. He has written "Drupal 7 Module Development," "Mastering OpenLDAP," "Learning Drupal 6," "Managing and Customizing OpenCMS," "Building Website with OpenCMS," and "Drupal 6 JavaScript and jQuery."

I could go on about this guy, but I just want to let him tell you a little bit about himself, and then we'll start talking Docker.

Matt Butcher: All right, actually working on another book on "Go In The Cloud" right now. Actually, this is the first time I've talked about it in public. I'm really looking forward to that. I'll be writing that with Matt Farina.

I'm the Lead Cloud Engineer at Revolv. Revolv is a company doing what I'm most excited right now about which is Internet of Things stuff. Internet of Things. We're building a hub that can connect a whole bunch of devices together.

There is an app component. There are lots of connected devices. The part I'm responsible for is the cloud part. I've got this great team here. One of the coolest things about working for a startup like us is that we're kind of inventing stuff as we go.

That means that we have a certain amount of flexibility in the technologies that we adopt. We can do so very quickly. I feel like I work at the toy store. We just have new stuff going on all the time. We're excited about the Docker stuff that's been going on.

We're kind of getting our feet wet in some parts of it. We really dove in in other parts of it so I'm excited to talk about that. My background, I came to Revolv directly from HP Cloud, where I was steeped pretty deeply in OpenStack. I did a lot of the aPaaS and PaaS work while was there.

My first big run-in with containers was as the lead on the aPaaS project that we were building a container-based solution that ran on top of virtual machines.

I've got all this cloud background that has given me certain perspectives on some of the stuff that I find changing as I watch this industry moving at an incredibly nimble pace. I'm one happy person. [laughs] I'm in my happy place. [laughs]

Lucas: That's great. It's really interesting, because you've got the enterprise-y, HP background, yet you're at a startup, so you have both sides of the vision of how both enterprises and startups could use Docker.

Could you tell us a little bit about how you're using Docker today at Revolv?

Matt: A lot of our core services were still running on full on virtual machines. What we've been doing is looking at what the emerging trends are cloud-wide. Docker is, of course, at the top of the roadmap right now.

Everyone on my team has played around with half a dozen to a dozen different container-based projects, many of them centered on Docker. CoreOS is a big one. Dokku, or Dokku, however you say it, and of course, Deis. The Deis guys are down the street from us here.

The thing is, we're really excited about all these things, but we haven't committed to any one yet, because they've all been fending off people, saying, "We're still unstable, we're still unstable." But, we're playing around with all of those actively.

Where we've really jumped in, is that we are using Drone now. We've replaced all of the clouds, automated CI and continuous deployment stuff with Drone. That total migration took us probably about a week and a half.

We're thrilled with the way Drone is working out for us. To me, it's exactly what we wanted. We were working with Jenkins. Jenkins was a nightmare to get configured, to keep running, and carried an element of danger with it. You corrupt a slave, and you spend the afternoon rebuilding the slave from scratch.

Drone has turned out to be the solution to a lot of our problems. That's where we first got a chance at building our own Docker images that are doing exactly what we want. Hosting Docker servers and virtual machines in Amazon. We've had a lot of fun.

One of the things we're very seriously looking at right now, we're doing a lot of exploration into the new Elastic Beanstalk services that Amazon is offering. They're now allowing you to deploy Docker containers.

That, for us, looks like the right kind of middle move between having Docker and containerized applications on the back end, and pushing those out into production on the front end.

We are still exploring, we haven't made commitments on that yet and we're measuring our steps, because we'd hate to have front line services compromised because we chose poorly.

Lucas: Makes total sense. One of the things that I want to talk about is about a year ago now, you wrote a very interesting article titled "Why containers won't beat Virtual Machines." I'm very, very curious at can our containers going to beat Virtual Machines now? Are you a leader or you have same view or has it changed at all?

Matt: Yes. It was fun to read back over that article because at the time Docker was three months old maybe. Most of the containers solutions were very new. Security people were still running wild, this may not be terribly secure. The container trajectory was still kind of down here system, where the Virtual Machines trajectory was up here, right there. It had gained a lot of momentum. OpenStack was stable. Rackspace and HP and IBM and all these big companies had entered the market for long.

Amazon of course was running with everything. Google had just announced their computing services. To me what the Virtual Machines were representing at that time was kind of the clear path to a stable and scalable technology and containers were looking like a faddy kind of toy. I fully expected when I wrote that articles that a year later we would see people's interest in containerization kind of winning.

I'm looking at as something to do as maybe an alternative for running a heavier local development environment, but I really did not see it coming down the pipe.

What's come out in colorless and with the PaaS layers on top like Deis, Libchanon some of these awesome things that are coming out of this community.

Basically, the virtual machine is still going to have its place for sure, but containers are really starting to look like a compelling model for grouping together, clusters of services, application running with each component and a separate container.

It's becoming a very elegant and compelling model. A lot of the new configuration stuff that I'm seeing come out CoreOs...I keep saying that CoreOs because that's the one I've been evaluating most often but I'm seeing it with all kinds of projects here.

From the Dev Ops perspective, these kinds of containerized models are turning out to look like the right way of going. From the security stand point, it's actually starting to look really compelling that I can control and shape traffic between containers very careful.

From the reliability and resilience stand point, the ability to deploy a container and have an auto-configure into its environment in a far more elegant way in what we're seeing in the Virtual Machine level is really looking great. This is Dev Ops Can Asia, Dev Ops kingdom of glory or something like that.

We could actually run these things and not have to fret every time a NUC goes down, that something somewhere in there is going to break. A year ago Netflix was talking about Chaos Monkey and kind the trend I was hoping to see really catch on in a Virtual Machine layer was this idea that we would test continually for the resilience of the Virtual Machines.

I think to some extent that sort of happened but in the container world, that's when they become sort of a core value. This idea that a clustered container should be able to self-manage, and that's what I see a lot of these technologies driving towards. That I find very, very exciting. My complaints originally with the container model was it, as I saw then people were saying, "Look, containers are better because we can boot them faster."

When I wrote that article, I was thinking, "Yes, sure, you can boot something faster but does booting faster mean better over Dev Ops experienced. Does it mean developers are going to be more productive? Does it mean customers aren't going to experience downtime?" I did not have the foresight at the time to see what was going to happen in the container war. I'm kind of happy to announce [laughs] that I was wrong...

Lucas: [laughs].

Matt: ...that I think containers really are going to be kind of the trend and rightfully so because I feel like the people who are focused on containers and focused on building and Eco-system of containers are focusing on the right things.

Lucas: Yes. That makes a ton of sense. I think that a lot of your points that you came up with in the blog that this space in [inaudible 10:34] expensive upgrading our lab security.

These points are either being handled already or the other thing that for me has always stood out, I think that a lot of the discussion on container and containerization has focused on those kinds of aspects. The security aspects and I think there is some value there. For me the break out features of containers has always been the social collaborative aspect of containers.

The break out feature of Docker is the Docker Hub. It's GitHub for the Dev Ops community, for the Ops community to be able to collaborate together and almost start sharing knowledge. Being able to share the ideas of how to put systems together, distributed large scale systems at a collaborative level.

What Git did amazingly well is take the source control stuff that had been done for years with CSV and some version and all that and make it something globally collaborative and distributed and hook into that broader knowledge. The GitHub has really brought out some great thinking and great collaboration.

That's what I see in many ways being a corner stone of what makes Docker so amazing, is it creates a source control for Dev Ops. It creates a way for us to collaborate together and share container images collaboratively and aside from the security aspects, aside from the boot up speeds, that's something that I haven't seen anything touch before.

That's what excites me about Docker and containers.

Matt: Yes, and you got kind of the funny unit stereo type. That on the one hand you the developers who are learning from each other and sharing code and stuff like that and then you've got the operators in the C segment who are hiding in the corner with some kind esoteric knowledge the rest of us don't possess. That's the stereo type I think it's gone back a couple of decades.

What's interesting about Docker's whole concept with the Hub style workflow is that it may turn out that the only reason that C segments were ever characterized that is because it wasn't ever really a good venue for sharing knowledge in this kind of world.

When Juju started get popular started see that happen there, I think that this is really opening a new world for people to be able to collaborate on, exactly as you put it, collaborate in a space where collaboration just didn't ever seem to work before.

Lucas: We've talked a little bit about what you like about Docker. What do you not like about Docker?

Matt: What do I not like about Docker? A week ago I would've said the perpetual putting off of a one point, really this is my biggest concern.

That got solved, that's great. My biggest concern now really is that a lot of the tools around it are still really missing and immature. That's a problem that would be solved provided people continue to be interested in it, and I think they will. That has been so far high on the radar that I haven't even really looked too much deeper to see what the other big concerns are going to be.

The way my team works, our particular workflow has been centered around Docker files. It feels like every time I find something that we can't quite do and another version comes out and we seem to be a little bit further down the road and I feel the kinds of needs people like my team have are the kinds of needs that many teams have. I'm very optimistic that those small kind of things are going to get solved.

Lucas: On a big level, the stability and the kind of, we're not going to break things of 1.0 was something you guys needed. But from a technical level what do think was the biggest problem for real life adoption of Docker? For companies like yourself, for bigger enterprises, is there something technical what you think is kind of a sore thumb?

Matt: The biggest thing, the one thing that I'm really looking forward to someone solving for us is right now if I want to deploy Docker into a production environment, I am back to Virtual Machines if want one of the bigger clouds. Even with Amazon Elastic Beanstalk, Amazon in my mind is kind of the biggest player that's really...Although Google now too has just announced they are going to support Docker tenures.

The way they're supporting leaves me wondering, are we really getting all we can get out of containers? I feel there's big space for someone to really, really solve well the public cloud version of Docker Containers. We've got OpenStack, we've got Amazon, we've got Google, we've got all these solutions for Virtual Machines based funds, but it feels like we're still stuck at the PaaS layer if we want a fairly robust Docker environment.

Once we're at the PaaS layer, we're really one layer too high to do a lot of the Dev Ops stuff that a larger [inaudible 16:20] even a medium-sized installation would want to be able to accomplish. We backed off our usage of Paas style frameworks, because we wanted to step down a little bit and manage more effectively what we were doing with our resources, how we were compiling things, what the environment was like.

Docker provides us that but we can't deploy it the way we really want to deploy it into a public cloud. One of my biggest reservations, the first time someone on my team said, "Let's switch to Docker" was, I don't want to do that because that means that I have to manage Virtual Machines, and then I have to manage Docker Containers on top of Virtual Machines.

If there was a way to make it so I don't have to manage the Virtual Machines, ideally they wouldn't even be there or would be somewhere where I don't have to know about them. I think that's a big opportunity. It would be a company like Revolv, we're kind of looking. We are at the high level PaaS, we don't want to manage multiple tiers of infrastructure. We want what feels like the natural bottom layer of infrastructure and we don't really care what's hidden beneath that [laughs].

Lucas: Yes. This is exactly the sort of thing that CenturyLink and CenturyLink Cloud are thinking about, that's why I spend a lot of time in this space. As a Docker user, were you aware of the stuff going on in this week's announcement for libswarm and libchan, and some of the new open source projects. Are you appraised of what's going on? Do you understand what it is? Are you excited about it?

Matt: Like many servers from a distance, we've been looking at things to see how they pan out. Revolv has made a pretty big commitment to Go at this point, to the Go programming language. A lot of these new libraries, a lot of these new technologies, have libraries for Go.

Libcontainer, there's an example of something that where we might be able to build some really cool stuff, where before, the barrier of entry in building a container wrapper would have prevented someone like me from investing the time, because I have other things to accomplish.

I was pretty excited about libcontainer. I could be the only one, but I'm really excited about that one. As far as any kind of real time messaging and message queuing systems, Revolv, in doing any kind of Internet of Things stuff, real time interaction is a very important aspect of this.

For example, you've got light switches. You don't want to have outlets and light switches that take three to five seconds to turn on from when you press the button on your mobile app or whatever.

Seeing libchan, which I feel like is really going to start propagating the Go notion of channels at a network layer, it's an exciting technology. I'm looking forward to it, and looking forward to seeing how it matures.

Those two in particular, have bubbled up to the top of our radar. Libswarm, I'm still not sure what to make of it. I feel like a need to sit down, read about it for a while, see if it's something that's going to be usable for us, or if it's something that we'll let other people make usable for us.

Lucas: You've mentioned that you're evaluating right now, things like CoreOS and Deis and Dokku. Can you tell us how you're thinking about these projects? How far along have you evaluated them?

I know our audience is thinking about them as well. They probably haven't had as much experience as you have. How do you go about choosing which one's right for you? Can you talk more about that thinking process?

Matt: With Dokku, that was the first one I started playing around with. That was almost a year ago now. It was fairly early on in Dokku's...I think whenever Hacker News covered it. That one was fun to play with. At that time, I was coming off the aPaaS project, which is an application platform as a service that HD Cloud offers. We've been working with a fairly robust PaaS solution.

To play around with something that was, as the catchphrase goes, "A hundred lines of bash script," it was fun. It was fun to look at. I've played around with it two or three times since. I like the idea of the minimalist build-your-own PaaS.

With D-Zone a little while ago I wrote an article for their cloud report about how PaaS are evolving over the last two years. They'd gone from the mindset that the PaaS was going to be the monolith. Developers would set their code gently on the top, and everything in between would be taken care of.

That was the old view of it. There are some decent stacks that were built that way. It's turning out that developers don't find that to be terribly satisfying once they make it past the prototype stage. We took a good, hard look at Deis and Dokku, and some of the other mini-PaaS, build-your-own PaaS sorts of solutions.

I really think they're promising. I'm excited to watch Deis mature, because it's gone from...The first time I tried to install it, I think I spent hours just trying to get all the Ruby dependencies right. It's coming along really well. I think their decision to leverage a lot of what's going on in CoreOS has been fantastic.

Our plan here, is we check on it every two to three months and see how it looks. My guess is that in two to three months from now, when we download it and start looking at it, the question we're going to be asking, is it going to be, "How does this thing work? What exactly are its features?"

But, is this ready to meet our needs now? That said, since we started evaluating that, we backed off a little from the PaaS layer and started to roll our own in a little bit lower level on the stack.

That's where I think CoreOS is looking really good. I feel like if CoreOS matures to the point where we can reliably deliver a cluster of CoreOS servers out into the cloud, and then use Docker and deploy into Docker, we're going to hit the kind of technology that Revolv is shooting for right now, which is the ability to keep services segmented where they ought to be segmented.

But, still have a single common infrastructure that all the engineering team and all the DevOps people can feel comfortable using.

CoreOS, to me, represents, again, that step down. I'm looking for that layer right there. Somewhere between having to manage a bunch of virtual machines and then another layer on top, but short of running on top of the monolith where all I have to do is set the code on top and hope that everything down the stack auto configures for me.

Probably, the next time we go through our evaluation cycle, in two to three months, we will, again, take a look at the PaaS layer, but we will probably focus our attention on whatever is maturing out of that middle tier there, as a runtime for containers, a robust scalable runtime for containers.

Lucas: That makes a lot of sense. We're actually working on a project at CenturyLink Labs right around that area that you might find interesting. We're going to be talking more about that soon.

You've been blogging about using Drone for CI/CD. Drone is a Docker-based CI/CD. It's written in Go as well, and it's like a Daemon. It's all packaged into one little thing. It's really easy to install. It's really cool, and it lets you do Docker-based CI/CD stuff.

I'd love to hear from somebody using it in a real life scenario. Tell us about it. Tell us about how it's different than using something like Travis or Circle, or some other CI/CD solution, or Jenkins. I'd love to hear real life use case.

Matt: In a nutshell, Drone works on the principle that when you push something into your Git repository, it will pull out a copy of that, start up a new Docker image, then execute whatever you tell it you want to execute on that codebase in that container.

Contrast this with what I view now as the old school, which was the Jenkins school of doing CI, where even if you could wire up the first part, so that all your initial checkouts and putting the code onto the slave board, you are still running on the OS, on the slave's OS. They've worked, of course, to try and integrate some of the Docker stuff.

What's nice about this, is from the moment the Drone container spins up, we know what the state of that container is. It was exactly the state that we told it to spin up in when the Docker file finished.

We drop our code into this known stable container. We can create and manipulate and compile and do whatever we want. Then have it respond by pushing out our end result. Then it destroys the whole environment. The next time it spins up, we've got a brand new, pristine, fresh one in exactly the same state we expected it to be in. The cycle goes on.

Initially, I had looked at this as a way to handle some of our automated testing. We'd commit, it would spin up a copy, it would run the unit tests, it would shut itself down. Piece of cake. Had that working in no time.

I kept discovering more and more things that I could do. I can tell it, "I'm also dependent on having some other containers. I want my database in there."

Now I can spin up the new container, load up the schema from my database into a separate Postgres container, then run all my integration tests and make sure that all that stuff is working well, too.

We kept experimenting and trying new things. In the end, we ended up with a very robust workflow, where on a commit to our master branch, the Docker will spin up.

It spins up with our own custom Docker image. It runs a series of pre-flight checks. It checks all the dependencies and makes sure everything's at the right version. It sets up the database. It installs the schema. It starts running through the tests. If it finishes the test, then it starts the process of building a Debian package. It builds our Debian package and pushes it into our APT repo.

After that, it uses Packer, which will spin up a new AMI for us, build us a custom AMI based on the output of that. Bundle it all up and push that out to Amazon. Then start up another script that we wrote that will spin up virtual machines somewhere off in Amazon. We can then immediately start doing our development testing.

We got an end-to-end CI/CD in no time. This was, like I said, project from beginning to end took us about a week and a half. Granted, that's because my colleagues are brilliant and could jump in here and do a lot of the esoteric stuff, like building Debian packages very quickly. That was fun. It was interesting. It works. We're getting a lot of mileage out of it. That's exactly where I wanted to be.

Whereas, how we felt with the Jenkins system was that every time we had to do anything it was, "All right, I've got to log into Jenkins. I've got to figure out what plugin to use." It was a burden.

Whereas, with Drone we've been more impressed by how much we could accomplish, and how close we could get to our vision in so quick of a turnaround time.

As I've been talking, I've been talking about how what we hoped to do is move off of virtual machines and on to Docker images. Drone, of course, is going to set up us perfectly for that.

In our end run, we hope that after we build those Debian packages and push them out to our APT repo, we'll be able to start up some new containers, and have the containers run and build themselves off of a Docker file that says, "Hey. Grab these APT packages out of this repo and get them going."

What I think Drone has offered, that a lot of these other past platforms have not, has to do with the fact that you can instantly have a very stable, workable environment. You can tweak it to your desires. We install our own custom packages, so we don't have to keep installing at the beginning of every run.

At the same time, when I really screw something up, and destroy the Postgres database, or I really screw something up and wipe out Etsy password or something in a container, which I've never done, but I could. [laughs] I did one time overwrite/bin.

When you do stupid things like that, the test fails, the container goes away, and the next time you start back fresh. Nobody has to spend four hours rebuilding anything. That was the freeing thing that helped us dive in and experiment with a lot of this other stuff and get things done so quickly.

Lucas: Interesting. That's fascinating. Have you tried Shippable, who is also doing some Docker related, but their more hosted version of Docker related CI/CD?

Matt: Yeah. I looked at them quickly. Hosted wasn't something that we could do at the time, and may still be something we can't do at the time.

I do think that some of these other Docker-based CI/CD tools that are hosted are going to be very successful, because they can provide this kind of tooling for people without even the hassle it took us to spin up a VM and get this running in there. It was out of scope and out of our ability to do it there right now.

Lucas: Go seems to be one of the cool languages to build stuff in lately. You guys are working on stuff. Docker's built in Go. Drone's built in Go. Can you tell us about how you got into Go? You had been doing some triple stuff before, now it's Go. What you like about it, and what you not like about it?

Matt: My kind of longer term background is as a Java developer for 10 years, then I switched and did some PhD and some [inaudible 31:03] for a while. Then, when I joined Revolv I joined again as a Java developer, and we had an existing Java API server.

I have had good and bad experiences with Java, but more recently I felt like Java developing was piling on library upon library upon library, so that I could do something. Our application really was starting to feel like that. I have been playing around with Go even way back when I was at HP and over in my three time building level tools and things like that.

I was impressed with the dual facts that on one hand I can build up remarkable powerful application just using the core libraries. That was because they focus so much effort on getting the networking right and getting the structure of programs right.

On the other hand, the fact that it compiles well to fairly generic binaries that are small on size and use a very small amount of memory, meant that I can run them on lower powered machines, lower powered VMs than what was writing in Java, which took many, many times the resources.

Those two things were compiling enough, but then once I really got going and realized...Rich Hickey gave a talk a couple of years ago, strangely that was called "Simple Made Easy." He focuses a lot on what consensual simplicity means. While I know he is thinking mainly about functional programing, to me Go really represented the allegiance of simplicity.

Not everything may be easy, but the stuff that's in the language feels like the stuffs that are to be in a language and the stuff that's not in the language largely feels like stuff I can comfortably live without. Now this is the sweet spot I really warranted in programing in general.

I think that's really a sweet spot all of us, right. We don't want a language where we are continually trying to learn what the core features are nor that we want a language where we have to download 60 or 70 external libraries before we can build something like a simple functional HTTP server. That got us going.

The funny thing is originally I was just building a demo that was going to power a little thing in a booth at a trade show. We invented our Cloud service in a matter of weeks, and went, "Hey, I want to keep doing this." It's stuck and our team pivoted very, very well, very gracefully from Java to Go ,and we are all just kind of 100 percent Go as far as what we are planning for the future Cloud projects right now.

Shouldn't say that, not a 100 percent. We have a healthy dose of side scripting languages like Python and Ruby and stuff that for our core services we really pretty much committed to Go over this point. Not because we had a do it by fiat, but because the engineering team feels like it is the rare language for building these kinds of tools.

Lucas: I really appreciate your time and I am super curious to figure out what you guys land on in terms of that middle layer, what orchestration you pick. I would love to check in with you in two or three months and see where things stand, if that's OK.

Matt: Yeah, that would be fantastic, that would be fine.

Lucas: Great, thank you so much for you time.

Matt: Yeah, thank you. It's been great.