Tech Labor Market – Deepak Giridharagopal, Puppet by Perforce

This is Textron TV. Hey everyone, welcome back to text strong TV. I'm happy to be joined by you know, what?

I've been interviewing him for a really long time. Probably more than longer than I want to. Admit to though he'll admit to it.

I want to introduce you to my friend Deepak, Gary. Hot ah darn it here. Polls close got the middle wrong the middle of those right?

I know it's it's a it's a spicy Scrabble score for sure. It's impact help me carry the gopal. So like really darapa.

Yeah, they're gonna be Dara popo. Yeah, you know what it is over the years Gary. I can't Deepak.

I can't read as well as I used to because it's very much is pronounced the way it's written it is so first started doing these I used to be able to read your name. That's a little fuzzy for me. I know right like well, I've been doing this a long time.

Yeah 10 years from now, we'll check in again and we'll finally get it. So well, probably having an AI or something a point. That's true.

It'll just be fake of me actually. Exactly, we won't even need to be around anymore. So exactly we'll be chilling at the metaverse.

I hope so try to stop. Anyway, they're all kidding aside Deepak is CPO of puppet, which is now puppet by perforce. You've been CTO there since what 2011 Deepak?

Yeah. It's been around in a minute for sure. You a long time and you know, I always enjoy getting together with tea pack because he has a great kind of view of where he sits and what the world's like and we're gonna talk about some of that today.

But before we even jump into it the first thing I want to Kind of get out is that puppet once again is conducting their annual state of devops survey. And this is sort of the Granddad of devops surveys. I don't even know how many is it?

12 years 11 years. Yeah. I think it's been I think it's been if not over a decade closer to a decade than not.

Yeah. Sure. So yeah the survey I think just recently open maybe in the last week or so for people who want to contribute to this year survey and I encourage all of you to do it Deepak.

Where can they go? Yeah, you know probably the easiest thing which just beats if you just Google I mean it's kind of a known quantity at this point. If you literally just Google for state of devops report, you'll find it.

There's also information about that on the puppet blog and then you know, if we want we can supply the actual URL. Yeah, I have it in notes will put it in in the show notes here. But if you watching this if you go to puppet Comm slash blog, you'll see yeah a Blog article up there on yeah, 20 20 devops.

Yes. So this one, you know every year we try to tackle a slightly different kind of thematic area in relation to to devops and the one that we're looking at this year is really around platform engineering Sort of as an organizational concept like how many people are doing it is that effective, you know, just tell us what your experiences are like, you know, good bad ugly all of that stuff. So yeah, you know just like in years past me encourage everyone to to fill out the survey like it's a great source of data.

Like we're not the only company that uses it like, you know, there's a the research I think ends up being pretty formative for a lot of like, you know, people's thesis on like new tools new organizational structures that kind of stuff. So yeah, that's a long time. Yeah.

Absolutely. It's one of my favorite things to seeing all the results come in just because it Usually every year it actually shakes up some assumption I have about the industry and there's so much kind of fashion and woo going on in the world of software in general. So it's nice to get some quantitative support one way or another right?

Absolutely, you know. It's funny. You mentioned platform engineering write an article.

It was a sponsored article not on our site. Yeah, a couple weeks ago devops is dead platform engineering kind of taking its place and yeah. Yeah, I read that you yeah, I I wrote in not bad to that.

I'm so tired of hearing that kind of nonsense. Right? First of Technologies.

Don't die. They evolve right? Yeah platform engineering is cool.

SRE is called devops is cool too. com. I think March of 2014.

Yeah. At that time, you know puppet along with one or two other companies. They were devops right when you said devops to people that's what people Focused in on yeah, right and and looking back on it now though.

We called the devops. It really was much more Ops than Dev. I think when we got the stuff with puppet and configuration management and you know deployment and stuff like that and you know, the world's changed since 2014.

That's puppet as a result and you know part of the world changes the the people the labor force and the skills required of changed and and we've seen the rise of things like sdsre. We've seen the rise of the platform engineer and platform engineering Doesn't necessarily mean devops has died. It doesn't necessarily what it means is you know these things.

They take on lives of their own and you know, they're pushed and pulled by market forces. Yeah. No doubt that.

Yeah. No, I I mean I broadly I would agree. I mean I think you know, one of the points we've always tried to get across is that you know, it's devops is not exactly people will quibble all day long.

Like what is the precise definition? You know, it's not like Webster's has, you know, a definition of this anywhere like a lot of things in technology, right like so so I think initially I think well some stuff is Evergreen, right? So I think the idea of having more of a shared set of interests and incentives that align a development team and a set of operators so that they could better come together to actually do what nominally.

I think they're paid to do which is deliver a service that users like that actually works and stays up and no one loses any more sleep than is absolutely necessary to keep the stuff up and running. I think if you go back to like 2013-2014, I think the reason why there was such centricity of tools in that conversation was really for a couple of reasons. I think number one operations as a field had a lot to learn from the development ecosystem, you know, like a lot of us were working in environments.

It's like why can't operators, you know represent like representing infrastructure as code was like a radical concept, you know for many people back then what do you mean? Why would I do that? That's what devs do and it's like no there's lessons there that make sense.

Like revision control makes sense. Like peer review makes sense, like all these practices we could inject but the but that was never the goal. The goal was never to turn operators into developers.

That's more a means to the end where the end is can we collaborate better? So we're not considered these like two foreign entities that kind of look at each other with suspicion because that was really the state of how things were. So if I fast forward to now, I mean obviously, you know is devops did I happen to have a pretty biased take on that?

But I also think I'm right which is I don't think it is. I think it's absolutely every single company of any meaningful size that I talk to continues to have this discussion about how to kind of get a more efficient more collaborative. You going right like everybody still continues to deal with this but I do agree.

There has been an evolution the tool chain underneath has changed the nature of infrastructure for a lot of companies has changed what skills are available in the labor market has also changed in some cases for the good and some cases that complicates things just probably a set of questions unto itself we could get into but I think it's never a bad idea to step back and reassess it but you know, that's part of the reason why we wanted to do the survey on platform engineering is you know, there's definitely a lot of interest in that but part of me says like well, How are people actually implementing the stuff and is does it actually carry out or fulfill the promise of what we were trying to get at in aligning development and operations and I guess we'll see when we actually get the quantitative data. But but ultimately I think the problems are still there like Services still need to be maintained people still need to be on call developers by and large still I think many of them don't enjoy being on call. They don't want to be in an operations role a lot of systems administrators are operators.

Don't necessarily want to spend all their day learning 12 different JavaScript toolkits to like do hardcore development some of them do and that's great. But I think as long as those preferences exist, we will have this kind of cultural divide, right and I think a lot of the sound bites I think presume a lot of an organization's capability of doing what someone writes you know in a blog post like if you just organize your teams in this way, it's all better and I think in reality what we Scene is it's there's really no one size fits all it's really more of a conversation at a backdrop than it is a prescriptive set of methodologies, you know, you know. Again, we've learned so Deepak you and I can talking for years.

We've learned so many we've learned so much and gotten wiser so much over the years. around this whole issue that you're talking about sort of what is devops and all of these things and couple of things first of all The world is such a complicated chaotic ball of mess sometimes right? It would be so good.

If you were starting with a blank piece of paper always and said, okay, this is how I'm gonna organize. Yes, my devs of my Ops. Am I this my platform engineering and how I'm going to do things?

And these are the people I'm gonna hire in there. Yeah. These are the product.

That's a fiction, you know one gets, you know, it's react to them. It's beyond rare, right? The fact of the matter is most of us working in brown messes of balls right round, you know, not Green Fields Brownfield muddy Brown fields, and there's technical debt and there's people and personalities and cultures that we have to deal with that.

We just can't You know snap our fingers or wiggle our nose and change their culture. Yeah, or their experience or their skills for that matter. Well for sure upskilling's hard, it's it's tough.

And I think I think you you raise a really good point, which is that what works for you know, three people in a garage and you know a Bay Area startup like I think about some of the startups that I advise right and usually one of the things I try to remind them is like you don't you don't have to do things in ways that scale because you don't have a scale problem yet, right. If anything that's an advantage that a lot of these smaller, you know more Nimble or startups have right and they need that to overcome all the other disadvantages. They have of being a new business but you're right for most people that are in an established environment.

Like why does Brownfield stuff even exist it exists because there's technology in play that it's actually doing stuff of value, right? If it wasn't valuable just just nuke it from orbit right why it's still there and people have Nation or they're at least cautious or methodical about how they want to deal with it is because it's usually fulfilling an extremely important purpose. Right?

Like it's the the transaction Clearinghouse for a stock exchange. Right? Like why isn't this written in you know, the latest version of golang or something like that?

It's because well rewriting it puts potentially trillions of dollars worth of transactions every day at risk, is that worth it? So I think the risk calculation is very different. So a lot of the models that we've seen that get a lot of hype, you know, particularly from like Fang and you know those types of companies, I think I don't know that they always translate to all these different like it's not I'm not Netflix where I have basically one product more or less, you know, and it's a web delivered product that can optimize things in a certain way that it's great.

You know, don't get me wrong. Like I'm not trying to minimize they're accomplishments from an operational standpoint, but that Playbook doesn't is not necessarily universally applicable right across like all the different. Insanity that exists in like it that was that was one of the early I think lessons is you know, I don't want to be Google.

I don't want to be Amazon. Yeah, I'm me I Don't know for sure. Not a unicorn.

I'm not from Silicon Valley. Kind of crazy. You know it just I think you've seen that in what I kind of lovely called this trend sort of the like the let them eat Cloud Trend which was like well Look the real way to basically emulate the efficiencies of what those companies are doing, even though I'm not like running a search engine where I'm basically printing money in my basement off a bad words or whatever, right?

But I'm an insurance company in the midwest. Like, you know, what I should really do is I should copy everything that they do if we just move everything to the cloud we'll be able to have like totally programmable infrastructure and I think what's happened and I think what the state of that is at this point because I think it's very much an attitude of like well if we just let all of our indeed eat as much Cloud as they want, you know, like how big of an Ops problem are we really gonna have because the cloud takes care of a lot of that when in reality, I think those of us that have been through at least one major Trend and downfall, you know and infrastructure like we know like all of these things come with trade-offs. So I think now what you see is you have either a totally kind of fragmented infrastructure set up inside of these it departments where you know, everyone's sure everyone's deploying on the same Cloud but nothing looks the same you have other things where you need to apply uniformity like for auditing or security purposes.

Like how do you do that? When everything is fragmented and you've kind of empowered all these individual teams and isolation, how do you hire people that have the skills to not only architect a new system architect it in the cloud having awareness of like distributed systems and Cloud native infrastructure like You're kind of looking for an extremely rare combination of skill sets. And if you're Google or your Apple or something like that sure you can afford to pay an extremely high salary for you know those sres but If you're you know in Middle America or something like that where cost of living is radically different the labor Market's different like you don't necessarily get to do that same you can't apply that same solution you may become afford it and even if you could maybe it's not a wise investment relative to other stuff that you can do.

So, I think you still have this exact same problem, which is like now, it's great that there's better infrastructure Primitives out there, but they're complicated to their own ways. So you still end up with I think a lot of the same problems we had in 2013 2014. It's just moved around.

I think the complexity has just moved around. I don't know that it's an absolute terms come down all that much which is a little disappointed very fun to mental kind of more physics. There's two kinds of people in the world right one kind of person things that the Universe continues to evolve.

It becomes more complex, you know, the camp says the universe involves. It becomes less complex. To get simpler.

Yeah. From the more complex world, right? And I think we've built things and we can do more but we didn't want some things may have been simplified.

They give rise to other complexities and yeah. 300% like I think about the I came to talk about Dynamics in here somewhere. No, it's a entropy right is kind of what I think about it and you even see that in the lingo like, you know, a lot of Ops folks or myself included we'll talk about like antifragile anti-entropy systems, you know control loops which are anti-entropy like the whole reason that those exist is they're based on this assumption that like look chaos is probably inevitable like lets you its own devices this stuff, you know will kind of rocks Natural State.

Yeah, even if it doesn't quite make sense why it is like eventually it will it's more likely than not or at least it's likely enough. You have to think about it ahead of time. But like, you know, I gave this talk a while back on like where was the origin of config?

I was kind of curious like what's the origin of managing configuration? Right? Like okay, you know, it's puppet when you configuration management orchestration, like when did that start and like one of the earliest references?

Fired was from a US military like Department of Defense like manual like a training manual and it was for people operating the control panels for missile silos, and it talked about the importance of understanding and logging the configuration of every switch on that panel anytime. You made a change you had to be meticulous about it. Why?

Because chaos can happen and you know, someone leaves a switch up when it should be down. Someone didn't write down why somebody did this and the one thing you don't want to do is accidentally launch a missile, right? But I think about it now probably the one thing you don't want to do is accidentally take down a stock exchange or accidentally, you know, take down a you know, the health insurance websites or things like that like it's equally important but what is really changed, you know, we still have those problems today.

The technology is shifted underneath like the Advent of I don't know RPMs or Debs is packaging. It's like that eliminate the need to understand the operability of a system and to that eliminate systems Administration. No, you have been a virtual machines eliminate that no one could argue.

It made it more important because now you can have more things I'll go even more basically the Advent of devops and as far away and platform engineering and every continue with testing and continuous delivery. Need any jobs absolutely not no more job. It created a lot because I think it you know, there's definitely there's the people out there that view this is maybe interesting given the general economic climate.

I will say and why I don't look at the stock prices of anything anymore because it just depresses me, but like Oh, well automation is the Refuge for these. To like reduce their head count and I think about this and I'm like, you know, even though the tools have become way more advanced. It's not that I still think like I've always thought it's really unwise to think about the tooling is Advanced now so I could turn my operators.

My sis had been my slres. I can basically think of them as rumbas, you know, if I just had sufficiently Advanced tools. I just don't don't say that.

Well, we'll get in trouble. I could I could turn them into automated thing, right, you know an automated a lot more efficient by yeah and instead I view it. I always thought that was short sighting.

So I think the real probe the real move the real strategic move is not to look at them is like I could turn you into a robot or replace you with a robot but instead think about it. Like I can build you and Iron Man suit, you know amplify what you can do. I just and we're out of time, so I'm not going to be able to have you back on but A lesson we've learned through all of this is don't underestimate the innovativeness and resiliency of humans.

Oh, yeah, and given better tools and giving you know. More resources they come up with crazy stuff that works I'll end with because we're out of time. I'll end with like I think I truly believe that complexity kind of expands to fill up how much space you give it.

So as you get better and they let you do fancier things what happens is people aren't like. Oh now I can simplify everything and do less in less time with less expense. Now people just do more and a larger scale quicker.

And you know kind of like the story of my hard drives anyway. Back you back on because I wanted to talk about lab. Or um, we do this three days always we always have to come back up.

Let's get let's get you rescheduled that will continue the conversation 100% for now though. I want to leave the audience. Hey, go take the puppet survey, please.

com blog check out the article with the link. Go take this survey. You'll help yourself.

And the whole industry is you always do it this deep back my friend. It was great seeing you. I'll have you on soon.

We'll continue. Fantastic. All right, good seeing you as always Ellen.

All right this challenge we're gonna take a break here on Tech strong. We'll be right back.