Ep#89 The Inside look at Empowering Developers

Host: Jon

Please join me in welcoming senior developer advocate at AWS. Cobus Bernard. Did the show Cobus oh my God. My cheeks are, we didn't even get started.

Guest: Cobus

<laugh> I haven't even started with the dead jokes yet. That's the worst spot. Yeah. <laugh>

Host: Jon

This is too cool. All right. Today we're talking about Terraform content creating and a balance of AWS services. And does AWS sunset services? I don't know if we're gonna answer that question fully here on the show because that's a very touchy subject. All right. So now a moment from our sponsor for some devilishly good coffee. Yes. You know, I'm talking about the diabolical coffee brand, my only coffee brand that keeps me going for that early morning, late afternoon, or just about any one of my favorites is a Columbia. It's a medium roast, but it's ly good. Oh, and did you know that you can get an additional 10% off on anything? Yes. Including this awesome swag, they got some cool stuff and some cool merchandise immediately available to you. Take a look at the description below for 10% off your order. Now it's time to head back to our episode, but first I think I'm gonna grab myself a cup of devilishly good coffee. Let's get to it. We might avoid it in general, but we do have some topics for that. First of all, cups, let's get a little backstory on yourself and let everybody know that you're in Seattle after a very challenging trip during the holidays. All right, let's get to it. <laugh>

Guest: Cobus

Um, okay. Let's, let's start with that. Let's call it a fun little adventure. So effectively. Um, as part of my team move, uh, internal tos, I'd also decided to relocate to Seattle goes, uh, just a lot of things made sense with the teams here and recording studios, availability, um, et cetera. So what I thought was going to be a simple process, you apply for your visa. Um, AWS has, uh, companies that help you with that. You go to the embassy, you get your visa, you book your plane ticket back up your house and you come over. That's how it's supposed to work. Except there was this little thing that got in the way called COVID. Um, so pro tip don't try and move during the middle of COVID when multiple countries are on lockdown. Um, so then what happened is obviously it was hard getting an appointment at the embassy in the, uh, in South Africa to go for the visa appointment because there was still closed. So Mauritius actually opened up. So we flew there, uh, for a few days, nice little like Beachside island holiday, which was quite a lot of fun, but still managed to get an appointment there caught in there, then went back to South Africa in November last year. Um, then it was also supposed to be very simple. We, we had booked the flight for middle, December to fly from Cape town, Amsterdam, Seattle, all booked paid no issues. No worries. Then, um, <laugh> the government

Host: Jon

Now the fund begins.

Guest: Cobus

Yeah. Now, now it gets like, and it's hilarious, like series of, or what do we call a comedy of errors, which is the first one being is that's around the time when this south African government announced the, um, uh, Aron variant, but then two days later, they also published data where previously we had about, let's say 800 cases a day, new cases. And then suddenly two days later we had a spike of 18,000 new cases. So the whole world went like, holy crap, hell no, let's not let people fly out the country. So effectively, a lot of the countries went to lockdown and said, you can't fly there directly. Now the funny part there is that it was just all data and the way the systems worked is that there's no way for them to historically add that into the old ones. So they just said, well, let's do, uh, summation and just pop 'em in as a single day.

Guest: Cobus

Obviously not a good thing, cuz guess what? This was at the end of November, I think on the 24th, this happened, we supposed to fly mid-December so clearly we can't get to the us cuz you have to be outside of the country for 14 days. So we're like, okay, no worries. Let's go to Kenya. It's one of the closest countries. We don't need a visa for it. So it's fairly simple. So we to go, go to Kenya for 14 days and we fly a little bit earlier, which pushed out the, or pushed up the date for packing the house, which was fun. Um, <laugh> we're still unpacking boxes and we're like, why did we ship this crap? We should have just gotten rid of it. <laugh> um, but yeah, some container joke to be made here as well. Um, get to Kenya, but on bomb, that's all I wanna come <laugh> yeah.

Guest: Cobus

Get to Kenya. Everything's on schedule, no problems until the day we fly. So we get to the airport and it's a flight at that point. It was Nairobi Amsterdam, Seattle, once again, very nice and simple. So we, um, try to book in and they struggling me with a system and we kind of like, we don't understand what the issue is. Um, and it turns out at the same time Seattle was having a blizzard. So the flight from Amsterdam, Seattle had been canceled. They didn't tell us this. Um, but they say, okay, cool, listen here your tickets to get to, um, Amsterdam, no, wait that one was canceled. The, the, the Amster and flight was canceled. So they really rooted us to Paris at that point. So we were like, don't worry. Get to Paris. They'll get you issues, uh, issue your ticket to Seattle from there.

Guest: Cobus

Don't worry. So obviously at this point, I'm like a little bit like something's going on airlines, then do this. I've flown a lot. You get your tickets. If everything is kosher. So let's go ask around and start looking. And, and I see that flight has already been canceled. So I'm like, there's no way we're gonna get there. So I thought, obviously figuring out an alternative flight. So I find one that goes Paris, Vancouver, Seattle, which is great. So that's fine. We get to Paris. They tell us, sorry, you can't do that because remember you need a visa for, um, Canada, because I'm from to Africa. I'm like, okay, cool. How do I do that? No, no, don't worry. This is online form. And take extend minutes. Just fill it out quickly. I try to fill it out. Sorry. Okay. South Africa's on a list of countries that cannot do the, um, electronic visa.

Guest: Cobus

You have to do a written one that takes three months. I'm like, oh crap. So finally Delta desk, um, support person helps me and finds a flight to LA once again with a don't worry, we'll get you to LA and then to LA will get you to Seattle. We've got a ticket by the time you landed sorted, get on the plane, get to LA, try and catch the next flight. It shows up not on the Delta app for a second, but only on the KLM app based on my, uh, loyalty program. So I'm like, I see there's a flight. Go ask, how do we get tickets? And they're like, oh no, no, no, that flight's canceled. And by, by the way, we can't see your ticket. So I'm like, okay, fine. When's the next one? So we book it for, this was Sunday. So we book it for the Wednesday morning, Tuesday night, they canceled that one. So we book for Friday, Thursday, they canceled that one. So we finally book on Saturday and arrive in Seattle on the 1st of January of the spending a week in LA with just our carry on luggage with two young kids, two and four.

Host: Jon

Okay. I don't really know where to start with all this. Uh, when did you start your journey and how long did you? I think

Guest: Cobus

We, we, we left South Africa, 11 December and we were arrived in Seattle one January.

Host: Jon

Wow. Um, definitely a challenging experience, especially over the holidays. No less. And with everybody else traveling and all their restrictions. Oh, okay. Wow. I think we should end this show on that. No, I'm just kidding. I mean, wow.

Guest: Cobus

<laugh> now that everyone's depressed about trying to fly

Host: Jon

<laugh> well, they've gotten a lot better. I'll tell you that. No, I can't say no issues. I haven't had any, I've been doing a number of flights and I haven't, but I didn't go from country to country to country. And I don't know how challenging that is, but I can imagine not only for you and your wife, but with the kids as well, everybody survived. Everybody's doing good. And now you got an awesome story to tell. Oh

Guest: Cobus

Yeah. Including like how to drive in the wrong side of the road, which according to the rest of the world is the right side.

Host: Jon

Well, it is the right. Right.

Guest: Cobus

But it's the wrong, according to me, cuz the left side is the right side.

Host: Jon

I don't think you're right.

Guest: Cobus

<laugh> Nope.

Host: Jon

<laugh> okay. Okay. So go, let's talk about a developer advocate at AWS. And then I wanna jump into Terraform cuz I think they lead into each other and uh, also some of the work that you're doing, what do you do? I mean, what is a developer advocate? I've had so many on my show, the stories range from everything, but the character really speaks for themselves and what they do and how involved they are in the community.

Guest: Cobus

Yeah. So effectively my role changed cuz initially I was the developer advocate for Sub-Saharan Africa and I think host county was 57 countries. Uh, which as you can imagine is an interesting one to try and cover. So my approach there was to see how much I could do, uh, online and in digital form for people to consume. But then that's in personal, you get very little feedback often with that. So I also did a lot of, um, in-person events. Uh, so I went to Ghana, Kenya, Nigeria, Zambia, I think, uh, Nigeria went twice, um, as well just for local conferences and some events. And then while there I would do meetups with things cuz the aspect with developer relations is that it's not a one way conversation or a, a broadcast system. We just go like, here's the info, use this where it's a conversation because you need to understand what are the developers and other technical builder trying to accomplish and do.

Guest: Cobus

And how do they approach it? Uh, because one of the interesting things with, um, Africa is that if you look at, for example, let's say a SaaS product or subscription service, um, that's, let's say 50 or a hundred bucks a month. That in us terms is like for company it's like, eh, that's, that's not too bad, uh, for a week currency. That's like, okay, no, we're gonna build something ourselves because this is way too expensive. So the whole approach is a lot different. When you look at that then obviously also in regional areas, they've got different ideas about how to build systems. So you need to take that into account. So it's a case of getting that feedback and then figuring out like how do you show people to build an important thing? There is to build the trust with them because we are not sales reps, we are not, um, what do you call it?

Guest: Cobus

Uh, uh, gold against how many sales leads that we create. We really, really hate that approach. So our, what we wanna do is we wanna show people how to build on AWS or just technical solutions of some kind and then help them in their journey and understand what their use cases are, how they go about it. Um, and then also as people are working on, <inaudible> gather that feedback and feed that back into the service team saying, listen, we've got a bunch of people saying this that isn't clear, or when working with this one, there's this use case that we just can't do with the service we have to like do this whole work around to address it,

Host: Jon

Having a, any sales or leads tied to a developer advocate is definitely the wrong approach. If you ever find a role that has that walk away because here's what's happening and I'm not only speaking for the developer advocate role, there are roles that require that that's fine, but you are building trust. You're talking with the community and the customer feedback and having that trust is key because they trust your knowledge. They trust your understanding that you have their best interests to not only show 'em the right path, but educate them on some of the things and bring that information back. If you're tied to certain sales, what's ultimately happen, it's really a self thing you're trying to do the best for yourself and it's not customer obsessed.

Guest: Cobus

Yeah, no, that, that, that's very true. And the one part you touched on there is the technical experience and background portion of that. So, uh, for myself, I can, I started off as a normal backend developer using Java and Linux um, then flipped over to the complete opposite, which is C shop and windows then flipped back to Java. Then we did like a fun, little roller coaster of mobile and starting to like do assess admin work, managing firewalls and file services and things that work while I did programming and then ultimately landed on what's now called DevOps. Um, and then turn to consulting for that, um, to help companies basically onboard to AWS automate do things faster. Um, and part of that poll long journey was like seeing these things that people are interested in, not interested and want to do and what are the common problems they face. Um, and that's also kinda like when my old Terraform journey started,

Host: Jon

Oh, Terraform, before I get this Terraform, I've got a check question for you just to get started Linux then went to windows. Why, wait, why, who does that? First of all, which one do you prefer?

Guest: Cobus

It was actually even worse than that. Cuz I came from at only known windows in my life. I knew about Linux systems. I played about around with them a little bit, but in no way, did I actually use it on a daily basis or knew how to run it. And then the company I was working at was very heavy on Linux and we actually used gen two. Um, and for those not familiar with gen two, that's the one where part of the build system done is the source code. You have to set the build flags, um, and manage like package dependencies, if there are problems and then compile it and then run it. So I went from, I think it was windows. I wanna say windows 98 at that point. Um, or windows, uh, windows XB went from windows XB to trying to run a desktop on gen two. And I still remember when K four came out, I was like, okay, I'm gonna do this right now. I'm gonna take the first release of a major release without patches. Let's build this. It took me I think, two days to compile everything. And then another week just before I could see my desktop, um, which was very interesting.

Host: Jon

Okay. I thought you were gonna say you went from windows 98 to me. That was an

Guest: Cobus

No, no, no, no, no, no, no. We, we don't talk about any

Host: Jon

<laugh> I owned it. I had it once gateway computers are okay. We won't get <laugh> uh, gen two it's really sounds like a Preble to, um, containers to Kubernetes. You had to download and pile the packages and all those things that you needed in order to run. Hmm. I wonder where they got the idea. All right. Let's touch on Terraform. Uh, why Terraform honestly, what is the value of it? Not only for you to jump into it, but what is the value you are seeing for others utilizing it?

Guest: Cobus

Yeah. Um, the biggest value there for the, the road that I've traveled is that Terraform isn't just for managing infrastructures code for your big cloud providers, cuz they split it into their credit provider for AWS, a provider for the other big clouds. They also have provider for other, um, services, for example, get up. She, so I had played around with, uh, firstly, just a bunch of bash scripts to set up route 53 cheap as I think in 2011 is when I started on AWS systems. It worked great until I had to make a change and I'm like, oh crap. Now I have to go edit the script, but I can't reuse the script and that just didn't work. So I started looking around what are other alternatives? Um, and then started playing with cloud formation in I think 2012 ish, somewhere big blob adjacent at that point I'm gonna blame cloud formation. Cause obviously I was young enough and it was definitely not me for putting every single resource in a single stack. Can't be me. Um, yes I have

Host: Jon

Ton. Yeah. It was simple. 1000 lines. No, I'm just kidding.

Guest: Cobus

Oh yeah. And then you hit the 51 KB limit at that point in up now figure out like how the hell do I split this, uh, without breaking production. So any case that was like, obviously my brain doesn't work with this. So I moved on to Ansible for infrastructure creation that actually went quite well for about four or five months until I got to a point once again, everything single stack, single, uh, role. And I started making typos and Python and Python has this lovely thing with Ansible. If you don't watch carefully, you would see this like big purple message pop pass that says undefined variable, but it happily continues. So I may or may not have broken production. Uh, I will not confirm that, um, with that and it, and I tried to fix it so I'm not okay, fine. Let's fix it. Then three days later I did the same thing and it's like, I got to a point where like I keep breaking things.

Guest: Cobus

This is just not working with, has to be something else. So I started looking in Terraform and for me personally, Terraform just clicked. It's just the way my brain works, where you define your resources, it figures out, you know, what the Delta is between what you want, what the previous state was and what it currently seizes the world in whatever system you're using. Um, and it was also great because I'd implemented a, a chef for conflict management and server provisioning at that point and then figure out, you can use multiple providers. So our developers are part of our get up org. They have individual profiles, they can manage their own public SSH keys. I can pull from there with their reform and then push it into the chef service. And the chef server would then deployed on our EC two instances for SSH access to the BA tip. So that was kinda like, this is quite neat. Um, and it just worked, like I said, the fact that you could do a preview, uh, using the plan function and actually see what you're gonna change. And it would warn you, I'm going to tear down this database that was super useful. Trust me. Um, it's like one of those, like I haven't dropped a database since not that I've ever done that. I promise

Host: Jon

<laugh> the mishap to production, not once. Oh yeah. Twice. Uh you've um, probably learned or sounds like kind of a DevOps has kind of come into prospect there where you're utilizing and doing some test ahead of time. And in fact it sounds like Terraform has allowed you to integrate more of a dev type culture where they can see the entire plan that walk through. I love the speed, the flexibility of Terraform and that you don't need to know your provider or it's kind of in a, uh, generic thing, the variables or implant, you don't have to rewrite everything. Oh yeah. Just to utilize it and spin it up or have to call out every single yeah. Little step. I mean, you could take a thousand lines of code and trim it down to like 200 and still get the same thing done.

Guest: Cobus

Yeah. And then you start building modules, as you figure out the patterns for things you add, like combine things. So at, at one, um, company I was helping, uh, with doing consulting, we would set up a module for, um, individual services. They were using containers. So it would provision the whole build pipeline for the container. Um, push it into ECR, then there's a deployment part of it. Then it would hook in Datadog for the monitoring added to the relevant dashboards. Uh, make sure the chefs are there, et cetera. So I just found that like that's super powerful. Everything is code. You can shield the developers from this by providing these templates to them and they're happy. And then when you wanna upgrade, you focus on the module and what needs to change and then just slowly roll it out, uh, through the different, uh, environments.

Host: Jon

I like the all, all the additional add-ons that they put out there and integration, GitHub, Docker, like they thought of all right, Hey, you know what? You need a plugin, you need a plugin, let's get all these set up so that you can do your full, uh, infrastructure code deployment without having to worry and write your own code that says, okay, let's pull the API for this and then integrate it into it, into the whole, uh, structure. Mm-hmm <affirmative> for somebody looking to get into Terraform. Right. And understand it. What's your recommendations?

Guest: Cobus

Well, I can do a little, uh, self-promotion there. I did do five videos on it. Um, on YouTube, uh, a while back. I still it's like on, on my top of my list of content I wanna create again. And yes, I'll give a link for it. It's literally just short ones giving you an overview and getting, helping you get started. Um, and I've done like a talk at Hashi Con sorry. Hashi talks, uh, original one in Africa about how to set up a multi environment. One, um, where the approach there that I see online often is people tell you to build like your infrastructure and modules. And then you have like either a repo per environment or you have a folder put environment. And that then relies on you copy pasting as you let's say, uh, promote different infrastructure changes to different environments, which once again, there are humans involved in this and I know myself, so, and I wouldn't trust me to do that, which means someone is gonna make a mistake at some point.

Guest: Cobus

Uh, so my approach, there was a little bit different where I, because you've got the variables inside term from that you can REU utilize, you can actually externalize them into variable files. So what I would do is have a variable file per environment, uh, that you then specify with the CLI, um, which end fuel, uh, environment you're dealing with. Uh, and then effectively environments are exactly the same. The only thing that changes is the configuration of the infrastructure. So do I spin up three web service for production or one for development? And do I go for like a large database of production versus small one or large one that's also high availability and replicated, and those are all just then variables that you set on an environment level, uh, which was very, very useful and powerful until I got to the point where I had to start working with workspaces, cuz then someone did the dev environment with a production workspace.

Guest: Cobus

And that led to a little bit of fun. Um, but as with all problems, nothing you can't solve with a bash script or a make file. So bolt a little make file that then would you specify the environment you'd even with, as a variable to the make file. And it automatically ties the variable file to the, uh, workspace and Terraform and that kind of solved it at least until somewhat I've upgraded the Terraform version, how to deal with that. So I built in MD five brute force checks for what binary are you dealing with? Um, and if you don't have the right one, sorry, I will not run.

Host: Jon

Oh, so you bought in some error checking. Uh I'd like it I'd like the integration for it. Here's a challenging question for you. AWS has CloudFormation, right? Yes. How are you? Uh, senior DA talking about Terraform and why not CloudFormation?

Guest: Cobus

Um, it's what I use the most. It's one of those things where use the tool that works for you. And we're not gonna tell you, listen, a lot of people use Terraform. You tthathal not use it because you are using AWS and, uh, uh, my friend and uh, effectively brother from another mother Darko. Uh, cuz we look the same. You should definitely get him on as well.

Host: Jon

I wait a second. You're not watching my podcast. I already had him on man.

Guest: Cobus

<laugh> oh, I have awesome. Sorry. It's I'm not traveling a lot at the moment. School commute is five minutes, so I don't get to do podcast anymore. It really sucks.

Host: Jon

Ah, it's all good.

Guest: Cobus

Um, but in case DACA's got this fun one where, um, cuz we did a stream, uh, for a couple of months together where his favorite answer is. If someone asks you, should you use cloud formation, CDK or Terraform, he would just go, yes

Host: Jon

<laugh> I like his answer. Uh, that's pretty good. And that's true. Darko. In fact I had him on there. We were actually did a quick video and a recording at the DC summit with Jackie and he had on the cool Jeff Bezos shirt. Oh yes. <laugh> uh, I love his humor. Yes. Having the two of you, you know what I think the two of you on a show sometime this is gonna be entertaining.

Guest: Cobus

Ah, I'm not sure we could do that. Cuz remember that'll prove that we are actually not the same person.

Host: Jon

Uh, I don't know how proof it is because it's still virtual. Right? I mean, I like, you know, it could be two separate recordings cut into it. Right, right. So it's not really proof. Yeah.

Guest: Cobus

No, but I mean it's one of those like super musing things. Cuz we were both on the EA D team. He was focused on central Eastern Europe and I was obviously sub in Africa and people on our team. So you work with him on a daily basis would once or twice, not often they would confuse the names, which eh, sometimes happens, but that's fine. But the funniest part was they would confuse and attribute work to the wrong person. So they would ask me about DACA streaming stuff or say thanks to that video that I'm like, that's not me.

Host: Jon

<laugh> how awkward and confused. Well I'm glad, but you both are in Seattle now. Yeah.

Guest: Cobus

He also moved to the north America team, uh end of lawsuit. So he arrived here in November. I got here in December cuz you know, it's you have to follow your heart. You can't let them escape.

Host: Jon

Okay, wait a second. So you guys are following each other. You're in the same state. You're in the same area. So you might be one in the same. I don't all

Guest: Cobus

Right.

Host: Jon

I cannot confirm or deny. Right? Let's talk about, let's talk about content creating because you create a lot of content for Terraform, for AWS. Let's talk about your process of content creating and how you go about

Guest: Cobus

Oh, that, that one is, uh, is, is a fun one. Cuz my role has actually shifted a little bit where currently I'm focusing on the content strategy. So it's like, how do we go about creating content at scale and dealing with that? Um, and I think the first part obviously is, and the most important one is start with a use case and what are people trying to do? So figure out like what it is you wanna show them or teach them. Don't just go, Hey, I've got a great demo to do X, Y, and Z. Now how do I get people to watch? Because that's the wrong approach. The whole point here is it ties back into the whole da role and building touch with people is like, you need to speak to them like a fellow human and effectively show them what it is they're trying to do or something cool that you know how to do.

Guest: Cobus

And, and also explain why it solves a problem. Cause once again, too often, you see, especially in startup, sometimes building a solution and then once they've got it built, they start looking for, okay, what's the problem. This is gonna solve and convincing people. There's a problem where it's so much easier. If you go like, Hey, I've dealt with struggling to get my infrastructure between the different environments up to speed and keep it up to date and make changes without breaking things. Now, when I come along and say, listen, I've got the school way with Terraform to show you how to build across multiple environments with just a pipeline to handle that. It's quick, the test, low level of error, then they're gonna go, oh, I wanna listen to this. Um, so that's normally the first step is figure out like, what does you wanna talk about? And it's like, what is the pain point?

Host: Jon

Okay. I got one word. Uh, or is it two words working backwards?

Guest: Cobus

Oh yeah.

Host: Jon

That's exactly what you're doing. Right? Yeah. So you're working back from the customers. In fact, I've worked for a number of startups and companies or, and they like, oh we got this awesome pro product. We need you to promote it. All right. Cool. Who are your customers? Who did it? So can I talk to the customer? Oh, we don't have them yet.

Guest: Cobus

<laugh> oh,

Host: Jon

Well I'm not promoting it because you're not get, you might get one person that's interested in the possibility. But do you ever noticed that when somebody built a product without being informed from the customer aspect, that when the product goes out, either it falls flat or customers are using it for what it wasn't intended to be used for?

Guest: Cobus

Oh, that, that, that second part is probably the, the most common one is like customers are exceptionally creative. Um, and they will find ways to use products. And I mean, I've done it in the pasta as well, where I use something completely, um, not intended for that. Um, I think the best example that I know currently of is, uh, Jeff, uh, Ling, he's very active in the raspberry pie and uh, community. And does a lot of videos of things is he actually uses, um, Ansible and templates and Ansible to generate his Wiki pages in, get up to link to his individual projects. Um, so effectively Ansible for building webpages, which I don't think was in the spec originally in the plan, but he does it and it works for him.

Host: Jon

Sounds very efficient. What about some other content creating aspects? Now you talked about, uh, how to grow and scale content in your new role. Right. But you're still creating content, but you have more of that education and knowledge on who's consuming the content. Yeah. So how do you go about it?

Guest: Cobus

Very, very slowly and very difficultly because, um, it's, it's hard, especially when you throw that at scale one. If I have to operate in a, like a, on my own without taking anyone else's contact into consideration, it's fairly simple. I can go work on what are the things I know a lot about and figure out what the use cases are for people out there and what they're interested in. So I would go and, you know, create videos and Terraform. Cause I saw people were interested in that. I can go and do a webinar on how to build containerized, microservice, architectures decide how I wanna split that up, which services to use, um, because I like them, um, and create the content very simple. It gets a lot more complicated when you start thinking like how does this content landscape look at, um, over, let's say the properties we have, we sort of have to think about, there's already this content that does exactly that with ECS, for example, but here's another one that shows you how to do that with the eeks.

Guest: Cobus

Oh, by the way, here's another one that shows you how to do that with just easy two instances and how do we then figure out like telling people, this is how you can do it, but then giving them the guidance. So that's, that's the more important part that I wanna try and focus on at the moment is like, cuz there's so much, but, uh, this, or have you thought about that? And uh, the classic dev answer to any question is, well it depends, um, building that nuance and what compounds that problem is. I've been at this now for, I think almost 20 years. So I've got a lot of historic knowledge and I can't remember how I learned, well, I can remember obviously the, the sequence of events, but if I have to go now with the large landscape of, if you look how open source is exploded with all the different, uh, just for example, JavaScript, front end frameworks and other projects and approaches and build systems and things you kind of go like, how does someone who knows nothing, get into this and start building that overall view of what are the components.

Guest: Cobus

And then once you have that view, draw into specific ones to know what they can do and then deeper even being able to understand the trade offs and figure those out. Um, so yeah, I'll let you know when I one day figure that out. Um, and then retire very rich.

Host: Jon

<laugh> I, I can see this, you, uh, the EC two ECS EKS, right? So you have all these things and you're like, well, you can do it. Well, it depends. I, I love that answer by the way. Yeah. It's a very classic kind of answer. Well, depends on what you're trying to achieve. That's an actual, honest question or answer, but how are you kind of, uh, displaying it to 'em but at the same time, are you empowering developers to come up with their own ideas? Like, Hey, here's how we use it, but here's a suggestion and then let them run with it. And the reason I'm asking that is because are you seeing more companies basically show how things are done, but showing some capabilities and now letting them run with it, like the open source market or letting them build and add onto it.

Guest: Cobus

Um, that's definitely a good way of doing it. Cuz part of that challenge is let's say a company has decided to do it a one way and it actually hits a sweet spot of perfect balance between complexity and solving the problem, which is kind of what you're aiming for. Others are gonna look at that. We have technic logic and say, you know what, that's a great approach. Let's start do ticket. So that then it starts propagating through the industry. Obviously there's a delay in any kind of new information in industry, how people do things. Um, and then it becomes very hard if you let's say decide that that's not the right way of doing it to actually go and change that. So having them run it's better than to figure out like how do I work with them or with that content and showcase some extra like gotchas or age cases, what ifs that people have to do.

Guest: Cobus

It kind like strange steer people to slightly different direction. If, if that's where we try and do, um, just to show the medicine, this is great up until this point, but thereafter you're gonna have the following pains, um, to help guide them with that. But I mean, I think ultimately it's effectively what works, um, until it breaks and that's so many people learn. So I'd love to say it's um, like I'm an expert in many things and I'm an expert cuz I can tell you all the ways to break it. <laugh> um, and it's, it's also actually one of my favorite interview questions is like if I checked with a developer or DevOps person admin, it's like I asked him, Hey listen, tell me about the last time something went wrong. Didn't work the way it was supposed to. And it broke something. And how do you go about figuring out what was wrong and fix it because typically if you can't tell me how you fixed it or found the problem, it kind of indicates you may not have actually worked on that and someone else fixed it. Um, whereas if you tell me, well, well, you know, I was using X, Y, and Z. I found on the docs that that part was wrong or I'd implement this incorrectly. And then I fixed it that way that tells you, okay, I get a good idea of your technical skills and capabilities.

Host: Jon

Oh actually I like that interview question. I know we're jumping on another topic of interview questions <laugh> but I, so I think I'm gonna keep that and I wanna write that down because I like that a question. The reason I do is because you, you tell me if you're going to learn and be curious, right. And I hate to use the Amazon term. No, I don't hate to use it. I love it. Oh, so you right, because here's what happens as a developer, as somebody who likes technology, you're gonna jump in there and oh yeah, you are a as a technical person. Alright. I gotta, I gotta switch gears. I gotta ask you a question. Just do ahead. How many times do you read the documents before trying something out?

Guest: Cobus

Um, only if the CLI tool or auto complete of something in my assistant doesn't work and I can't find a name that matches normally. Um, it's like typical dev. I go in like head first building, see how it works. And then only when something starts breaking, it's like, Hmm, what's what's going on here now.

Host: Jon

That's exactly. Every anybody that is hands on to developer it, that likes technology, you will go in and play around with it and break something. There's your story for your interview question? Well, to be honest with you, I, um, yeah, I just went to the command line and just hit Ron and I took out production. Well, I learned that I should read the document first

Guest: Cobus

Or be careful what you select. So when I started my career,

Host: Jon

I, this

Guest: Cobus

Is gonna be good. This is a good one. So I did, um, electrical engineering got to like third, third or fourth year. I can remember the mix of subjects I had finished at that point, but just, I just want to code. Um, but it was one of the things where, um, my grandfather's actually helping pay for the studies and he were still old school. You know, what are these new computer fangled things? That's just a phase. That's not important. You need a proper career like engineering. And I just wanna code. So effectively dropped out, uh, went, just finished off a diploma at, uh, a local college, which did do coding and things like databases, except I learned SQL with paper and a pencil. Sorry, no pen. Cause you went lot pencil. Cause you went around to raise your answers in the page and yet to do the invitation in everything on paper.

Guest: Cobus

So that's how I started, you know, I'm prepared for, you know, this coding career. So day one was here's your computer, set it up, get it going, figure out, look at the docs and things. Day two, I was given an interesting problem cuz you know, I know SQL, which is we've got a database of about 500,000 users. Uh, we've got some kind of sign up bonus, um, that they can use for things. But we are seeing that people are, um, abusing this. So I spent about until about two, three o'clock in the afternoon, writing this exquisite sequel script, the way clause was like this, figuring out like if you see the, this pattern in the emails or if it's that kind of number formula and the names and everything, and I'd run the select multiple times, um, to make sure that these are the right ones that I wanna, you know, lock the accounts for perfect came to, you know, updating it. So change the top from select to update, um, and ran it. And I sat there for a second because firstly, it ran really quickly, a lot faster than the select, which was the first warning. Cause our place tend to be slower. The second one was like the output at the bottom of, uh, SQL servers. Um, uh, I can't remember what the studio is called. Um, was 500 something thousand rows affected and I'm like sitting there going like the select that's 30,000, um, um, oh shit.

Host: Jon

Did you update your resume after this

Guest: Cobus

<laugh> no, no, no. So the best part is I had locked out every single user in the database that way. So it wasn't obviously immediately ping the senior day that was, um, uh, assigned to me and I was like, Hey listen, how do I fix this? So it was a case of, okay, cool. Let's, you know, sort out priorities, unblock everyone, cuz we already had the signup bonus problem. So just unlock everyone and now let's figure out how we can start passing old logs and look at loss activity and those kind of things. And that's kind of like where I learned like, you know, these, this log thing and you know, uh, an audit of events and actions that were taken in the system is kind of important. So yeah, I didn't get fired, which is great. Um, but yeah, it was, it was definitely a very fun day too. Uh

Host: Jon

<laugh> I can see this. That is actually a good interview. Answer to your question. Here's what I learned. One don't select start. No, I'm just kidding. Uh,

Guest: Cobus

No, no, no, no, no, no, no. The first lesson is don't give the new hire that has never done programming production access on day two. <laugh> I think that's the core reason

Host: Jon

<laugh> well actually nowadays in DevOps, that's pretty much what you do because you have an extensive pipeline for testing and everything. But if you're just on day two and don't know the processes. Yeah. I, I think you might wanna hold back a,

Guest: Cobus

That, that little button there of like you have a testing pipeline in place is a fun one because I don't know if you know, but did you know that every single company with a product out there actually has got official testers for those systems? Some are just lucky that it isn't their customers.

Host: Jon

Oh <laugh> it's like a Mike drop I'm out.

Guest: Cobus

No. So I can remember where the quote comes from. It's uh, it's definitely something, um, I almost wanna say is charity majors from a talk you did in 2013 around, um, I think about that time where I heard that one's like, that's like when you start, the penny drops, like, yes, we need to obviously check before our customers cuz the last thing you want is the customer to come back and say, listen, this is broken. And that's how you learn,

Host: Jon

Uh, customer obsession going back to content, creating. What do you find challenging during the process? Cause we're gonna get some Amazon interview questions here at the end that I want to ask a little more specifics on. No, I'm not asking you an interview question. Actually, we should do a loop on the show now <laugh> don't worry. I'll bring on Dar. I'll be, you know,

Guest: Cobus

Okay. Could be quite fun. <laugh> what

Host: Jon

Do you, wow. That seems like another podcast for testing the loop process. All right. Anyway, we're getting off topic. What do you find challenging around creating content or basically fun? I mean you might find it as a challenge process, but it's one of the things that you enjoy.

Guest: Cobus

Um, typically I don't find content creation, um, difficult as such. Uh, the reason is like I only take on content that I've got experience in and secondly, I'm excited about, so there is some tech that I've used in the past that I don't think I will, you know, wanna be excited about or wanna do a talk about. So, and that comes across, like if I'm doing a talk on something that a I'm not familiar with or B I'm not excited about, the audience will know. Yeah. They will pick that up. So it has to be something where you can literally hear me start talking faster and faster and then going like, Ooh, and then you can do this and then you can do that. And then, um, and I have to look back at my sticky that says speak slower, um, which have my screen.

Host: Jon

So wait a second. Does that mean you're getting excited about talking at this podcast? Woohoo. Passionate check on <laugh>

Guest: Cobus

I love just geeking out and talking about tickets. One of the things that I think in my career, the initial part is I didn't do a lot of meet and things I didn't even know about meetups and anything about, you know, day to day work and the people you'd work off. There was obviously, and yes, I'm giving away my age. Now there was some IC channels where you could chat on and then later on we became friends with people. You had personal, I CQ message. You could share with them. Um, but yeah, there wouldn't a lot of

Host: Jon

People. There's some throwbacks.

Guest: Cobus

Oh yeah. I still, I don't have an active IQ anymore, but I still have my numbers somewhere. I'm actually check if it works,

Host: Jon

Uh, somebody's gonna bring it back. You know that all of us, uh, older folks who have that information still and saved it, it's coming back. I think you, you mentioned, uh, passionate about and knowledgeable on the subject. I think they go hand in hand because oh yeah. First of all, I I'm passionate about a lot of things, but knowledgeable on the subject, that's where the real passion comes out and you start to speak and you totally geek out. I, I have to, I have to diverse here a second and jump onto another topic only because one you've touched on the things I was on a pre-call for a person to join my recording and I should have clicked record 30 minutes. We were geeking out 30 minutes talking about each other's digital setups and recording and content recreating. And then they asked me like something I'm like, wait a second. I gotta show you, let me show you this. And they're like, oh my God. I was like, we, it was probably the best. And the, the conversation I had in a pre-recording, I mean, it was amazing if I would've clicked to record it, would've been a whole entire podcast.

Guest: Cobus

Oh yeah, no, it's it's you get those people and those are the kind of people you want to create content and also have those conversations with. Cause like I said, it's like, if you're not excited about the product, the content is going to be may. Um, and the second part I think of the that's important for content creation is to think about what content you wanna create that actually adds overall value to the existing content out there. Uh, cause what I see a lot at the moment is there are so many courses on getting started with X and then it's just the one thing. And then people take that and just regurgitate it a little bit and change it a little bit. Now, granted, there is a part there that says take what there is and improve it, uh, enough that people will want to consume it and then they will find it and it'll start ranking, but effectively it's, don't just create because you can create and you know, there's already a lot out there. Um, and that's why I did the Terraform stuff. Cause there were some videos and content out there, but I felt like it wasn't structured the right way. It was missing the point about, you know, why you're doing certain things, how to think about the little blocks, um, and how to grasp through that versus just, oh, here's how you set up everything step by step. You know, let's just go through the mechanics of talking about it.

Host: Jon

I think that's very helpful. Here's a question. What about, what if I run into a problem? So an example, you're trying to do something that has say not even Terraform, right? Uh, the other day I ran into an AWS SSO issue for the CLI and, and way back in the day used to have to actually go out there, get the key, set it as a variable within and then run your profile, call out the variable. It was a whole, and then AWS released like AWS config SSL, which is one of the best things I think they did. And I was like, oh my God, I gotta share this with everybody. And I put together a two to three minute video on sharing it. Nobody asked me for it. What about that type of video and content?

Guest: Cobus

Um, that is definitely good. I mean, if you see something that you struggle with, you can't find an answer. That's actually a great point to, with a caveat if you found a solution to it, um, that is <laugh> decent enough that others want to use it. Um, example of this is that I was in situation at some point in my career and be very vague here where we had deployment issues with windows. And the only way I could get around all of the restrictions imposed was to set up Jenkins. Um, and then, uh, the, uh, the worker nodes, I would install the agent as admin on the specific windows machines to actually then run the commands and deploy and do things to the machine instead of having some conflict management system. Cause they just wouldn't allow that, uh, I couldn't use chef cause they would not allow me to open up S ports to the machines or what I can remember what the windows equivalent of that was.

Guest: Cobus

Um, but yeah, so it's like, yeah, make sure it's a solution that you don't wanna teach, that it will get a lot of people, uh, into bad situations. Cuz I often see that too much is like, um, oh, you know, just, just, you know, set up SSH, then put a password on it, just use that or just, you know, store the ver uh, the API keys and the text file over here, you know, nothing will go wrong ever. Um, so I think that is that little caveat of when you think of a solution and it does solve the problem. Just take a quick step back and say, how can this potentially not be a good way of doing it? What can go wrong? Um, just because otherwise you perpetuate this this way of using things. So your SSO example is a very good one. I still see content out there that tell you as like, for example, a Terraform in your provider, conflict copy and page your API key inside the Terraform file. That's gonna get committed to your repo. And that's like, no, no, no, no, no, no, no. We've way moved way past that. No one should be doing anything like that anymore. Nowadays

Host: Jon

Almost like I am a roles and access keys. We should be using roles instead of access keys for it and attaching the role to the EC two instance. But the key back then that was all the way that you had to do it, but we've moved to a more security model. Yeah. And doing it and having your access keys in your get repo. You're just asking for problems.

Guest: Cobus

Oh yeah. If you really wanna have fun, go generate a pay. And obviously just so that the right length change them or revoke them and break them and just commit them and see what happens. You'll firstly, get a flood of like automated email systems telling you, Hey, this is in there. Some people will pitch your products saying we've got a vulnerability scanner or, um, same sort of information scanner. You wanna use this and I can guarantee you at some point, someone's gonna try and use that. Um, within seconds of committing it, uh, they've gotten so good with automation.

Host: Jon

Aw man. I just wanna test that out. <laugh> I just wanna test how many people are gonna go try it or put out like the C two instance and when they finally get access, it does nothing. Um, yeah, that would actually be a good episode. So folks today we're gonna post our key. We're gonna do it live and see how quickly we can get some reaction from them. Let's do it. Oh yeah,

Guest: Cobus

No, it's it is very interesting.

Host: Jon

Oh man. It's some tests. We gotta do some more podcasting ideas go. But let me ask you a couple more questions before we wrap things up and the questions are about Amazon interview processes. We're not gonna get to sunset E to BS services. Well, maybe another time, but for this time, we're gonna talk about <laugh>. I know we said we get to it, but this has been a very interesting conversation. And, and I think having folks know that Amazon interview process is key. You already touched on one of 'em that you like are, and I assume you're an interviewer. Yes. Uh, you do a number of those. What are some tips that you can give folks not to not only prepare for the interview, but even the loop process.

Guest: Cobus

Okay. Um, the big thing there is to understand basically how our process works. So Amazon as a whole is a datadriven company. And that means that our interview process reflects that. So what we're trying to do when we have the conversation with you is gather data points to substantiate a specific view. Um, a good example of this is that, um, I believe our interview process, we do reference checks to ensure that you did work. We say you worked just to confirm that, you know, you're not lying to us, but we don't ask for references. The reason for that is a reference is subjective. And it's not a solid data point. Cause for example, let's say DACA wants to go for a job and you know, he's a, he is not a good spot or something went wrong. He's he is by no means a bad person.

Guest: Cobus

He's actually very awesome at a job. So bad example maybe, but let's say he, wasn't a great person you're trying to, you know, help a buddy out. Um, you might give him a good reference even though, you know, he's not that good, which means that if we take that as a data point, it means that we can't use it because we don't know if it is accurate or not. So we don't even ask. Um, and the second part of, and I think this is the one that often trips people up is like, we've got a very specific format of asking the questions. Um, and we would like you to structure your answer and what's known as the star format, so effectively build out what happened. So you start off with what was the situation then? The second one is what were the tasks that were assigned to you?

Guest: Cobus

Um, then we go on about what actions did you take? And then ultimately, what was the result? Um, and the fun one there is that, remember, we're looking for data points around how you think, how you solve problems, how you approach things. Um, it doesn't have to be a positive example. So if something went wrong and you can tell us the final result was X, it was not a good one. There's likely gonna be a follow up question. So how do you approach it differently in future? Cause we all make mistakes. We're not expecting people to come in and say, listen, I group pristine record. I never dropped a database table in production. Um, I don't wanna talk about that type of scenario. Um, I actually used that one as one of my examples in one of the questions I had to do.

Host: Jon

How many times do people realize that's actually very true? <laugh>

Guest: Cobus

Oh

Host: Jon

Yeah, this is not a made up example. This is very true with validate.

Guest: Cobus

It's true. I mean, if you want me to tell about everything that I did wrong in my career, there is a very long list of things I'm made, um, who do differently. Um, I mean, I even got fired at one point. Um, so yeah, it's been an interesting career, but yeah, doing something, the, the big thing there, the nugget is, mistakes are not bad, but we do expect you to learn from them and you need to show us how you approach it differently. Explain that to us and, and show it. And also just remember it's while it's Amazon. Uh, and I had this when I interviewed like, oh, it's Amazon, it's that massive company. Oh crap. I mean, this is gonna be like, you know, we're all just people, it's a conversation. Ask questions back. Um, cuz that's one of the part that I actually love most about interviews is that at the end we go, okay, cool. What do you wanna know about Amazon? Let me tell you about it because I mean, I love working here. I really enjoy it. Obviously as with any job, there are some things I like some things I really dislike, but overall it's still a lot of fun, exciting and challenging. So ask me about those things. So we can have a chat about that to see if you are interested in, you know, what things excite you, what type of things you wanna do, et cetera.

Host: Jon

Gobi. I want to add on to that with your data points and an example that I used in a previous conversation since Amazon is a data-driven company. If you're going to say, and there's two parts to this one, if you're gonna say we it's fine, we all work in teams, but be specific on AI. If you, you're not comfortable talking about yourself, get comfortable because here's what you need to do. If you keep saying we, you can't differentiate between what you did as a team and what was your role in the entire project? And then I have one more point, but I'm gonna let you comment.

Guest: Cobus

No, but that one is so important. Cause that for me as well is very difficult. Cause I, my approach is like we work together as a team, we get things done. I'm not a glory hog. I don't want to, you know, be put on a <inaudible> like, Ooh, look at what he did, that kind of thing. And it's very hard for me to kind of like brag about the things that I did. And it feels like bragging when you try and the answer in that way. But in that case, be very specific and definitely clear up what it is you did in that project.

Host: Jon

The other part I wanna add and I'll use YouTube as an example, if I say like I'm a social influencer on YouTube, I have have so many views. It's awesome. I've increased in month over month. So what if it doesn't pass us? So what then it doesn't matter. There's no data point on it. Now if I told you that I'm a social influencer and in the last six months, I've increased up a hundred thousand views from 200 to 300,000 views and added 1500 subscribers with an average viewer duration of 17 minutes. Shit. Yeah. You know, that's data that, that is solid data. I can prove that data, I can provide you that data, but now you realize, okay, what were some of the lessons you learned and how you did that? What was the results or changes that you made? Well, here's what I did is I've changed my podcasting style to more of a natural conversation like we're having now. It is completely unscripted and I don't know what's gonna come outta his mouth or my mouth that's but we have a lot of fun, good time. And we educate folks, eh, entertain with the possibility of education.

Guest: Cobus

Oh yeah, no, that's a, that's an important aspect. That one I know DACA always has that conversation with people that want to get into live streaming is that it's a conversation. So we're not gonna get on stage and just broadcast to the people earn we happen to do as live stream. And the second part is like, it has to have some element of fun and entertainment in there. Um, you obviously wanna teach people about things, but if it is not enjoyable, why do you even bother? Yeah. There's so much content out there. I can easily go pick something with a person who's excited and they make jokes and it's fun and it's informal. So why would I go to a stuff webinar if I don't have to.

Host: Jon

Exactly. And, uh, webinars are a thing of the past. Podcasted a new way. You can still have an effective, I wanna say brand awareness or a podcast with even a slide or two is fine. I don't mind a slide or two with a premise, but it has to be a conversation and don't sell me something directly. Tell me why it's valuable. If you start trying to sell me an existing product or service. I mean like,

Guest: Cobus

Oh yeah,

Host: Jon

It's great.

Guest: Cobus

And the, I think that's an interesting point with dev. If you, uh, I've seen like just on Twitter and other places, the number of companies suddenly hiring for dev and spinning up departments. I think the penny is funny drop that guess what? Traditional marketing does not work on developers. Yep. Our day job literally is to sift through information and make, uh, judgment calls on what is and is not accurate and what it is. So we spot marketing a mile away. Um, so yeah, it just doesn't work.

Host: Jon

So Cobus traditional tech marketing doesn't spiel where it should be at with regards to Dell developers. And the reason I say that is because here's what happens. Marketers know their job. They know how to do top of funnel, middle funnel. They know how to get leads. They know how to do all the aspects of it. But think about where they're trying to gear it towards the C level folks, the people who are buying, but developers are the ones who are actually buying. They might not be signing it, but they're the ones using the product. And they're the ones that are going to say, this is what we need. Now, if the sea level does it end up buying something else, the developers might not using it. So now you're collecting dust you

Guest: Cobus

Or they leave. Well,

Host: Jon

Exactly.

Guest: Cobus

You believe about, uh, over technology choices.

Host: Jon

Yep, exactly. That's why in, uh, I'm gonna reward it as traditional tech marketing fails to engage their audience.

Guest: Cobus

Yeah.

Host: Jon

I mean, it's really key.

Guest: Cobus

It is. I mean, you, you literally have to, and I mean that point about the developers it's, it's becoming more and more common and you can see it in a lot of the like reports and studies being done is developers are increasing, um, the number of them making the technology choices at companies. If you look at year on year, um, stats, more and more, cuz what people are finding is that when the developers know what they're talking about, they know why they want to use that product and what it solves for them and why it makes things easier. And also if it's reliable or not versus the slick marketing pitch for product X and when you put it in, you realize like Uhuh, holy crap, it's horrible.

Host: Jon

Exactly. They're the ones who are actually gonna be utilizing the product. I've noticed in uptake in developer advocate and develop rail roles that are available. Now, this is not something that was happening six months ago. It a companies are realizing the community engagement and the involvement for it go is before I wrap things up, where are you gonna be next? When can we see you next in person? Maybe?

Guest: Cobus

Um, for the first time in three and a half years at being at AWS, I will be speaking at reinvent and I actually have a session on, um, Terra form and things.

Host: Jon

Terra foreman <laugh>. Is that really the title for it?

Guest: Cobus

No, no, no. The title title's a bit more fun than that, but uh, we're not at a point yet where those are publics. I have to be apparently tow the line. They tell me

Host: Jon

Now that's an interesting way to do it. Thank you. So for joining me, this has been awesome.

Guest: Cobus

Yeah. I really enjoyed this chat.

Host: Jon

Wait, did you, you didn't sound entertained.

Guest: Cobus

I don't know. Sorry. I just got a ping. I have be very rude and run to a doc read.

Host: Jon

Okay. Well hold on a second. Let me wrap things up. Uh, because that wasn't really exciting ending, man.

Guest: Cobus

No, no it wasn't sorry. Okay. Sorry. We, we can cut that part. I literally got the ping saying like, listen, the docs blocked. Let's go.

Host: Jon

Yeah. Okay. So Cobus thank you so much for joining me.

Guest: Cobus

No, no, thanks very much. This was a lot of fun and this is the kind of chat that I basically, if I could just have my days with this, talk about tech all day long and play with co that would be like, you know, my dream.

Host: Jon

Ah, awesome. Hopefully we'll have you back on the show with Darko, everybody. Senior developer advocate at AWS Cobus Bernardfolks. My name's Jon Meyer. Thank you for watching the Jon Myerpodcast. Don't forget to hit that light subscribe end notify. Because guess what? We're at it. <laugh>

Guest: Cobus

Ding binging.

Host: Jon

<laugh> yeah,

Guest: Cobus

Exactly. Cool.

Host: Jon

All man. I know you gotta go. Yeah.