Random Acts Of Automation

  |  Compiler Team   Automation and management

Compiler • • Random Acts Of Automation | Compiler

Random Acts Of Automation | Compiler

About the episode

Forget the high-stress fire drills and chaotic, late-night maintenance shifts. Not only does automation provide identical, predictable environments—it acts as a structured foundation for security and compliance, turning daunting, multi-step manual processes into reliable push-button tasks. But what happens when we combine automation and AI?

Global automation strategist Chad Ferman joins the show to demystify how automation and artificial intelligence can work together, rather than competing.

Compiler team Red Hat original show

Subscribe

Subscribe here:

Listen on Apple Podcasts Listen on Spotify Subscribe via RSS Feed

Transcript

We used to have every Tuesday evening, like, the people tasked with deploying the thing would all be staying late, and then QA had to stay late to check things after the fact, and if everything went down, you gotta deal with it. You got to stay up late for it? I had to go in at 4:00 in the morning. That's the benefit of West Coast hours. Yeah, I guess so. That was East Coast Tuesday mornings. Lady- Yeah ... East Coast Tuesday morning. Wow. Yeah. No fun. Mm-mm. Now that there is automation in place, like, nothing wrong with using automation to get away from that kind of work/life lack of balance. Yes, please. This is Compiler, an original podcast from Red Hat. I'm Emily Bock. And I'm Jennifer Scalf. On this show, we go beyond the buzzwords and jargon, and simplify tech topics. This season, we're covering the fundamentals of IT infrastructure. We're closing out this season. Today's episode is all about automation. You know that classic enterprise fire drill scenario? A catastrophic outage happens, everyone gets called into a room, it's intense. Sometimes you're there for hours, even days, talking in circles. Know it. I've lived it. I know, time and time again. I feel like actually in the middle of one of this season's episodes, I got called into a fire drill much like that, so. Oh my gosh. They're certainly still around. That's right. So we're, you know, like I mentioned, we're at the tail end of recording, and that was very close to the beginning, and it really stopped us in our tracks, right? And that's what happens. Oh, yeah. And that, like you drop everything because there is a giant fire, and that needs to be everyone's focus. And it's all, like everybody's holding their breath at the same time. Oh, yeah. They're all just sort of... And then who knows what's actually broken. Mm-hmm. Right? It broke. When somebody in IT says, "It broke"- Yeah ... which is the classic stereotypical case opener, right? Yep. The case is opened with, "It broke," or, "It's not working," or, "I can't reach it," or da, da, da, da, da, it. Um, yeah, y'all, I have been in IT support for a very long time, day two, and that, that's the feeling. Um, it broke, and everybody says, "What is it?" and, "What are we doing?" And who- everybody's holding their breath. Yeah. And we have to figure out what, what, um, logs do we need? What... We're going back to early episodes of this podcast of this season. That's exactly what it feels like. It's like putting out a fire, except you're not just- ... putting out the fire. You also have to figure out how the fire started or it will come back. Yeah. So you're, like simultaneously solving the mystery, fixing the problem, and usually dealing with a very upset customer or multiples. Well, I was about to go into the, the business criticality of it. Is it an internal situation? Is it external facing? Mm-hmm. How much money are we losing that this is down? Constantly- Yeah ...in my world. Like, what are the consequences? Like, there is a difference between, oh, this is annoying and I'm mad, and oh no, everything has ground to a screeching halt. Right. Am I, uh, mildly inconvenienced, or is, are we literally losing money because our customers can't get something done? Those war rooms are intense. They are never fun. They are, they are never fun. Why do we choose to do this last, Emily? So I think- We're gonna talk about how you fix Ugh. Okay, fair, fair that situation in the war room, or hopefully avoid ever having to deal with it in the first place. Oh, my goodness. Our nightmares. I'll bring us back a little bit, because I think last episode we talked a lot about security and compliance, and those are things that typically can be seen as really slow, really painful, but they're extremely necessary. Those steps are necessary, but they really shouldn't be reactionary either. Like, these are things that should be built in. Uh, don't wait until... We, what did we just... We're making up metaphors again, right? So the house is on fire. Yeah. Don't wait... This one's so easy. Don't wait until the house- ... is on fire to check your fire extinguisher. Okay, that one was way too easy. We'll have to come up with more creative ones. That's it, though. It's like there's a lot of things you can do to fireproof your house so that you never have to deal with the giant blaze. Yes. And if you build that in from the beginning, like in the architectural plans, et cetera, then, you know, you don't necessarily have to go through that fire drill at the very end. Or that scavenger hunt. I like... Oh, I just thought of that one too, right? So let's not just have everything on fire. Let's do something a little bit more fun. Yeah. There's something good at the end of a scavenger hunt. Exactly, but when you also have to piece together all the clues to figure out what went wrong, that's not something that's very fun to do in a high-stress- Yes slow time kind of situation. Exactly. And everybody take a deep breath, hold it, and let it out. Next. Yes. Fire drill over. Yes. Now we can proceed. And for the final part of our season about infrastructure, we wanted to know how automation can change that story, so we talked to someone who knows a lot about the topic. Chad Ferman I work with a lot on a day-to-day basis. He's a global automation strategist here at Red Hat, working specifically with Ansible, and Chad knows firsthand how an automated configuration can turn compliance into a breeze. The way that I actually got into the last role that I was in is I was at a company that got acquired and I, I passed an audit in, uh, 24 hours, which had never happened before, and it was because they walked in the door and said- Where's the au- where's the machine that we're gonna run the audit on? I was like, I, I don't, any machine you want. I don't care. And they're like, "No, no. Where's the, where's the machine? Where's the server?" And I was like, "You can do it on all. They're all configured the same." For some helpful context, Chad's former company had to endure an 80-step manual process just to spin up a single virtual machine. And this, I'm gonna nerd out in this episode, 'cause automation is my bread and butter, and this is where the magic happens. Right. I forgot to say at the top of it that I'm just interviewing you this time. This is your, this is your wheelhouse. Yes. Okay. So Emily, tell us, what do they mean by auditing? What, how do they have the same configuration on... Oh, my gosh. And why in the heck was there an 80 manual process? I'm so confused about so much. I have found that every process has started from some kind of manual process from scratch, and at some point, once it gets over, like, a 30-page process doc of manual steps to get through things, that's when people start going, "Hmm, I wonder if a computer could do this instead of me." And I think that's how a lot of it happens. Yeah. No, now that I'm thinking about it, I guess I've always come in a little bit later. I've been a little bit spoiled in my career in that sense, because, like- Yeah ... it sounds like what you're dealing with is folks that are really starting from the base, from the ground, which is, that's our audience, right? So yeah, tell me more. Yeah. Well, and I think, like, this day and age, automation is really well-known. It's i- baked into a lot of processes, so it's a lot less rare than it was, you know, 15, 20 years ago, where it was probably just a script that someone wrote and ran on their own computer, and it was kind of silent automation that was happening, but it wasn't done in really a comprehensive way. And a lot of that has changed over the years, 'cause my, my previous company, we had a very similar process. Lots of manual steps that you had to do a million times over and over and over again, and it ended up being, like, a bunch of teams copying and pasting things out of a big spreadsheet. You do one thing wrong, and the whole thing is awful. So automation is perfect for situations like that, where you're doing the exact same thing a million times exactly the same way. That's what computers are great at and what humans are not. And I I wish we were doing a, um, a video podcast at this point, because my, my... I just keep making faces at Emily. One of the things that I learned and was taught as a sysadmin 20 years ago was if you do something three times- Mm-hmm you then write a s- at the time, a small bash script, then we wrote small Perl scripts, then we wrote Python scripts, da, da, da, da, da, right? Yeah. To not do that ever again. Like, you were chastised if you're like, "Why are you doing the same thing again?" So we just come from really different backgrounds, and I kinda love it. Yeah, it makes me happy. Yeah. Well, and I think, like, I originated in, like, the operations land. Okay. Where we didn't necessarily have, like, the, the immediate jump to automation. We had to talk to an engineer for that to happen. Oh. Oh, hi. Hi, I'm a new engineer. Hi, yeah, okay, that makes sense. Exactly. So it had to become a real big problem before you got to go bother engineering about it. Wow. That is really neat. I'm learning so much about Emily today. Well, I think that's a little less common nowadays with lots more options around automation. It's a definitely more of a, a first pass kind of thing now versus back in the day when it was like, oh, now I have to spend time and money and resources to build this automation. It's kinda taken for granted a little bit more that there's value to it. Yeah, and I'm, I'm having a lot of sympathy for the folks that are listening thinking, "But I haven't gotten there yet." Don't, don't worry. Don't worry. That's okay. Yeah. This is the first step, knowing that it's an option. Yeah, more, um, information about this coming up, right? So... Exactly. And so I think there's a lot of reasons why people have not used automation if, like, they haven't really gotten into it. Some of it is just learning curve. Like, for any given automation, like, back in the day, you had to know how to script. Nowadays, you might have to be familiar with a tool or a product that can help automate things without that level of technical knowledge, things like Ansible, you know, self plug. Little biased, but it's okay. It's wonderful. Incredibly biased, but it does take time- To learn about the tools that are available, and then also think really holistically about what your company is doing, what your teams are doing, and what parts of those are right for automation. 'Cause not every process is good for it. If it needs a lot of nuance, a lot of judgment, it has to be different every time, automation is not your friend, because it will put everything, all the square pegs into the round holes, no matter what. I'm laughing. I really wish we were, we were video sometimes, because I just can't stop laughing at the, at the metaphors too. But also how different our, our experience in these times were, because I'm also thinking back to when, I was here when Red Hat purchased the company that was Ansible, and I was here. I wasn't biased towards Ansible. So Emily comes in, she's biased towards it. I'm so biased. I wasn't. It's true. I've used all kinds of things. I won't mention what we used at my last company, 'cause nobody should use it. Not at Red Hat, at a different company. But I just wanna say, I'm not biased, and it is absolutely phenomenal. I mean, I love it. It is, it's my baby. But it's not the only option out there. You know, if it doesn't suit your needs, I, you won't hurt my feelings that bad. All right. Fair. But I think in terms of things that are good for automation, before I get too deep in the weeds, 'cause we talked all about, you know, containers and networking and all those things throughout the season, and anywhere where that kind of standardization, like a standard image or a standard process to spin something up, that is where automation kicks in. 'Cause it's a whole lot easier to push button, have thing go, than to do all of the steps required for it to happen. So Chad honed his skills pretty well before he came to Red Hat, and naturally he has some thoughts about the relationship between automation and, of course, AI. The way I look at it is automations are generally something that is programmable. So something that has been programmed to do something, which don't get me wrong, an agent will have that kind of logic and things built into it that it could say to the model. There's a lot more interpretation involved than when you have an automation. I can go run this Ansible playbook and know that it will set this system to do this thing this way every time, consistently. And it's that consistency that you get from an automation that you don't necessarily get from an agent. Chad brings up a couple of interesting points. Jennifer, I think we both know companies in 2026 are highly focused on managing AI token costs. It's been a nonstop conversation for at least, like, three or four months. I feel like before that it was just like, "Sure, tokens for everybody." Exactly. Everything keeps changing so fast, so. Oh, yeah, it was like a glorious two or three months where it was like, "Use all of the AI." And now it's like, hmm, maybe think about what things it's good for. Do you really need all those tokens? 'Cause we were subsidizing them. Oh, no. Exactly. And you know, automation is another tool that can help in reducing those costs. Yes, it certainly is. When you have it all scripted, as he mentioned ... Oh, well, okay, I think he said programmed, but I'm gonna keep saying scripts- Mm-hmm ... because that's what we called them back in the- it was always Bash scripts or Pearl scripts or some kind of- Oh, yeah but once you have that, you're not throwing it out there for an AI agent, model, et cetera, whatever you are using, I'm trying not to be specific- Mm-hmm ... to guess, because when it's guessing, I had a recent experience where I was talking to one of them. And it just kept asking me more and more questions, and I got to the point where I'm not trying to say this is some kind of conspiracy, but I swear it's asking me all these just to keep using them tokens, you know? Yeah. Is that just me? It definitely feels ... Like, whether or not it's true, it does feel that way, because you're trying to get an outcome, and when it keeps trying to ask you more questions, it's like, "Just give me the outcome." Just give me the out- And maybe that's part of my becoming a better prompt engineer, as the kids say. Sometimes I'm really good at it, sometimes I'm not. But if it is scripted, if it is programmed, if it is in some sort of automation platform, you're not gonna run into that. Yeah. A word that Chad used, declarative. Yeah. Oh- And I think that really is, is the difference for me of when I use which thing. Mm-hmm. Because AI, you can be like, "I wanna do this thing. I don't really know how it should be done, or there's options. I don't really care. Um, go forth and think. Give me options, et cetera." That AI is really great for. But anything declarative, meaning I declare you will go do this and then that and then this and then that- That is automation. So they work really well together so that you can have, you know, AI decide what to do and then set off the automation like a little Rube Goldberg machine. That's, that's the combination that's really powerful. Yeah, use your expensive tokens for the thinking through it step, and then when you know it, stop using all your... You don't have to use tokens anymore. Yeah. You can automate that and not use it. I like that so much. Declarative is such a good word for it. Ah, it's great. Yeah. It was like, that's what made it finally click- Mm-hmm ... kind of for me, of like when there was that first, you know, hype machine of- Mm-hmm ... of AI. It was like, "AI will replace everything," and I'm like- Mm "But I do automation. What, what do..." And so that's what made it click of like, no, there's still a home for automation here. Like, they work best together. And I have to say also this journey of the last few months that we've been recording together, we've all been trying to figure out what we're gonna do next, and it has been really good to figure this out, where we are still needed. Yeah. You know? So. Yeah. Like, I'm sure people are really sick of hearing about AI, but I know it's at least helped me to go through the season, be, be like, here are all these big Big topics, big categories of jobs and things. And like, yes, AI has a place in all of them, but it's almost never like a one-to-one replacement. Mm-hmm. It's an enhancement. Yes. It's a tool. It's a tool, and that's, that's my mantra lately. And my world, right? Like, these are really highly regulated industries that I'm in. And there's a lot of places where we can't have AI agents running wild, let me put it that way. Yeah. We've all heard horror stories recently. And so the compliance part of it, we were talking about regulations and compliance in a previous episode. In these industries, financial industry and tele- telecommunications, you can't have AI agents just going wild, right? Especially right now 'cause we're not really entirely sure what they're doing. Yeah. So there is definitely, I think, Emily, you, you've, you're safe for a while. I like to think so. And I have no power to prove that right now. But I think you're safe right now. It's funny that you mention financial regulations, 'cause I think that's the exact example that Chad used. Are you kidding me? 'Cause he works with a lot of financial institutions. Oh, rock on. Okay. Of course, being a technology company, we, we love all the tech and we're like, "Oh, MCP this and MCP that and this," and like, a lot of our customers are like, "Yeah, we have to set up an OIDC provider and make sure there's a security around every agent, and do all these different things before we can even talk about talking to an MCP server." Oh, yeah. So true. I know, I'm trying to hold back all of the horror stories that we've all heard now probably a thousand times, but if you haven't, go look, where somebody has unleashed an agent, has unleashed, um, ugh, one of the, a thousand other wrappers for agents onto their email, onto their- whatever they've got, right? And it's ruined things. So one of the things recently, and, uh, I guess it was last summer. I'm trying not to date myself too much, but last summer. In AI world, that's 20 years ago. I know, right? It was a big deal that folks were coming out with all these MCP servers. And so what I saw in telecommunications world, a lot of it... Yes, those are great. That's wonderful. I'm not gonna go into great detail what about what MCP stands for or what that is. Y'all can do that. But know that i- in your businesses, you need to have someone that is- Verifying and certifying them and whatever that means inside of your company. And then having a tokenized, but not like the agent tokens, but inside of your world, your login, your single sign-on. That's what he meant when he was saying OIDC. Yeah, yeah. Well, let's say that a couple more times. 'Cause folks need to know what this is. If you're in setting these up, these are-- this is all the new stuff that changes every week and a half also. So how do-- what does that mean for you to have, like, a whole bunch of different MCP servers for your folks to connect? And I'm trying not to say, like, the actual names of the products, because they keep getting bought, and their names keep getting changed, right? So I'm about to say one, and then it's gonna change its name. It'll be obsolete. Yeah, and when this, and this episode comes out. So we've got all of these s- these, these wrappers that connect to all these different AI agents. And then to bring in context, 'cause it's really m- okay, I'll just say it, Model Context Protocol. I, I think that's what it s- what Model Context Protocol. Okay, MCP. Um, yeah, yeah, yeah. Okay, fine. Um, but it literally makes it so that you don't have to keep telling these agents the same thing over and over again. That's a very high level. But you don't wanna just connect to any of them, because if- Yeah ... you're having a conversation with an AI agent, you've brought in a Model Context Protocol that's telling it different things. You don't know where that ca- if you don't know where that came from, if you haven't logged in, uh, in-- with your, your, uh, single sign-on, your company's credentials, et cetera, you could literally be pulling in the brain of some nefarious character. Like, I'm not trying to get too freaky on anybody, but this is where we're at right now. Yeah. And so th- that's a really good point, right? What he just said's a really good point. I think so, for sure, 'cause I kinda see it as, you know, connecting to an MCP server is not all that different from signing up for a new service from a vendor that maybe your company has not vetted. Oh, yes. Uh, maybe you're not familiar with. Like, it can be just as dangerous as that if it's not something that has been looked at and approved and, you know, considered safe. Same thing for agents. You don't want to give an agent Access or credentials or tokens that you wouldn't give an employee at your company. It is not that different from a security standpoint to having, like, a personal assistant. Like, you can give them access to things that you trust them with, but they then have access to it and can do whatever. Yeah, and I'm a little bit obsessed with the whole MCP thing right now. Again, I know I'm about a year out of date, but it's like bringing in a personal assistant, and then they bring in their cousin, who has a totally different thought pattern and experience and whatever, and that might be valid and good, but, like, we don't know that cousin, you know what I mean? Like- Yeah ... they could be adding stuff to documentation. They could be giving their opinion in meetings where they don't, they, I mean, they're wonderful people, but they don't have the experience that we need. It could be something that's not nefarious. It could just be extra data that's not relevant for what we're working on right now. It's really wild how these things work. Exactly. No, it, it introduces risk of some form, and I think much like how we handle, you know, vendors or people, the antidote to that is, you know, company policies and people checking those things. Those policies that, like, your HR or other various compliance teams put in practice, the equivalent on, like, the technical side for AI can be automation. It's a layer. Layers and la- layers upon layers. But yes, so if you have that AI has freedom, the automation can help both kinda predict what it will do, if they can work together, and it can also act as the guardrails around it so that if AI do this- say, "No, it don't." Yes. Oh my gosh, absolutely. I didn't actually know Ansible could do that, Emily. We need to get more deep into that. All right. Every tool is as powerful as you make it be, so- Yeah, that's cool ... the value there is anything you can dream up, you can probably build automation around. And, you know, there's a lot of tools to help you do that. Oh, so cool. And when we come back, we'll hear more from Chad and close out our exploration of modern infrastructure. All this after the break. Another reason why automation matters, well, if you do break something, you should be able to turn back the clock. You should never have automation that you don't have a rollback plan for. You want your automation to be automated to a point where you know that it is not going to break something, but even if it does, it has automated rollback. Thinking through, like, all of the different things that could happen to cause a failure, and automating around those so there is a fallback or failback to the desired state is super important when architecting and designing out good, good automation When everything is automated, teams can go back and check the logs to identify the issue and roll it back smoothly. Hopefully, anyway. This is why knowledge bases and uniform processes are important. Automation strategy isn't just telling every team, like, "Go automate things." That's really not how it works. Uh, you have to have, like, some kinda plan, and proper access controls and logging, and I w- uh, during one of the breaks, started to talk about a situation where I accidentally deleted a very important part of a password file a very long time ago. Oh, everybody has a story like that. Everybody has that, and will never forget, and will never do it again. Um, but there really wasn't any logging, and there really wasn't any way to log, roll back, so we had to do... This was so long ago, we had to, like, I think, put a USB th- or a CD. I mean, it might've been so long ago, we had a CD or DVD, so. I do think that everyone has a story like that. There, like, there's a, um, the industry legend of, I think it was Toy Story. They accidentally, like, deleted the whole movie and were saved by someone who had, like, taken a copy home to work on during maternity leave. Oh my gosh, that's amazing. Those kinds of things happen, and they happen a lot, and so I really like what Chad was saying there, which is you have to design your automation to work like you want it to, which means you need to design it to be logging every step, so you can go back and see what went wrong. You need to design it with rollback options so that you can recover that previous state. Wow. Good automation, like pervasive throughout your company, doesn't just happen. You do typically need to have a plan and then execute on it. My gosh. Yeah, so I only broke email for a while for some faculty members. Uh, we fixed it. Oh my Go- I'm laughing now, but I'm, the cold sweat, going back to the beginning where we were talking about what it feels like. Yeah. The cold sweat that you break, break out in, I mean, I, 'cause I care. You, I, if you guys have noticed- Mm-hmm ... I actually care, a lot, so I try to pretend. I try to be all cool, calm, and collected, but I do not wanna go back to those days before automation. That was insane. Yes, and definitely not automation without, like, a- Oops, undo plan. 'Cause automation is very good at doing a lot of things very quickly. Yeah. And that can be bad if you tell it to do the wrong things. And oh my gosh, I'm remembering the whole s- series that we talked about. Remember that one story I told where the automation kept overwriting and we could not figure out. Remember? Oh, oh, it all comes together right now. So we could, we, there was no oops, go back button in that situation because it just went and checked. It did the diff that Chad was talking about, and said, "Oh, hey, this is different, I'm gonna overwrite it." But it was like five minutes after the box had booted. Yeah. It was- And that's where the logging would come in hand of like, wait, what happened? Am I going crazy? Yep. Reliving it. I think it's also important for security and for updates. Like we talked in a episode earlier this season about updates and what happens if you get stuck in the middle. Do you roll back? Do you keep going? And like, I remember when working past midnight on patches and maintenance was a normal part of the job, especially for upgrades or deployments or things like that. Another, um, lifetime ago that I don't wanna go back to. Thank you. Yeah. And I think now that a lot of that is automated, you don't see that as much as you used to. Like, we used to have every Tuesday evening, like the people tasked with deploying the thing would all be staying late, and then QA had to stay late to check things after the fact, and if everything went down, you gotta deal with it. You got to stay up late for it? I had to go in at 4:00 in the morning. That's the benefit of West Coast hours. Yeah, I guess so. That was East Coast Tuesday mornings. Lady- Yeah ... East Coast Tuesday morning. Wow. Yeah. No fun. Mm-mm. Now that there is automation in place, like nothing wrong with using automation to get away from that kind of work/life lack of balance. Yes, please. And I think Chad would agree. The reality of it is, is if, if things are automated, if they're properly governed, and there's consistent methodologies around how you get from everything from how you observe a problem to how you automate and fix a problem, just like building that muscle memory where it just becomes something that everybody just does as part of their daily job, and there's a way to do it that's consistent across an entire company versus the, "Yeah, just make sure you're automating stuff." Well, I like what Chad is saying. Yeah, no, I think that's almost exactly like what we were talking about with compliance- Right ... in the last episode, is that like you can't just check it at the end, then it's just a checklist, and you haven't really adopted anything. But automation is a great way to literally build that checklist into your process because that is the process that it follows. Exactly. And yeah, you don't just check the box just to check the box. There's a reason that you're doing it. Mm-hmm. And there's a reason that we're automating what we're automating. I'm trying to think of any silly ex- I, I, again, I'm running out of steam on our silly, you know, examples or our, our metaphors or our analogies for that. But yeah, there is a reason that you're doing these things. There's a compliance, um, requirement. There's regulations. There, there's, each industry has their own. So go and automate those things that help you stay compliant. I think it's a great first target that you can grow from. 'Cause I think, I, I'm sure we're, we're not alone as a company in trying to develop our AI processes where it was just like, "Oh my God, everyone just go use AI. Figure out what happens, what works." Yeah. And coming at it with a plan- Ends up with a much better result than kind of the organic growth that comes out of just everybody kinda doing something. Yeah, and I'm over here again and still obsessing over those MCP servers. Like, don't, don't... You know, like ahead of time, find out what all of those components mean. There are many, many people out there now trying to figure out what each of this, these, um, the agents, the components, the, all the wrappers. I keep calling them wrappers 'cause I can't figure out a bigger, better name for it. And then- Yeah ... find out how do you automate how your users interact with that, right? Like, automate those single sign-ons, automate how they get access so they're not just ru- running around saying, "Oh, look, this is a new shiny tool this week, and this new shiny tool that week, and this..." You know? Like, keep them interested. Yeah. Keep it consistent, and you'll also save a lot of money. Everybody's not just throwing- Yeah ... all kinds of crazy stuff at different agents. Yeah, if you're already struggling with your token allowances- ... like, automation is a great way around that. Exactly. Cause if you can get it to be like, okay, if this happens, then set off this automation, then the AI doesn't have to do all of those steps by itself, which could be different every time. It could decide it wants to do something else. And it does. Or you can use less tokens to have it set off something already known and logged and vetted and all those good things. Exactly. Emily, I wanna ask you a thousand more questions about automation. I wanna answer a thousand more, but- I know, right? What, what did we miss? Like, there's something I know there's something in the back of your mind right now, like, we didn't have a clip for or whatever, but you just wanna make sure everybody knows, and I wanna drag it out of you. No, I'm just kidding. I wanna hear it from you. No, I know. It's like I know, I know we talked a, a lot about having, like, a plan for automation, and I know that that can be, especially if you're starting kind of from scratch from, like, an enterprise company, that can be really intimidating. So it's not that you have to come out with, like, a fully formed, totally thought out strategy that touches everything and everyone and launch it all at once. That's not necessarily what an automation plan means. Mm. What it means is being deliberate about how and where you introduce it. So iterative. I love the word iterative. Iterative, yeah. So you don't have to do everything all at once. This isn't, you know, launching a giant product waterfall style. It's a lot more like agile- Where you're delivering discrete bits of value at a time while making sure it still fits into the bigger picture, but not letting it just be kind of random acts of automation. Okay. And then bragging when each of those work, right? Let's get-we gotta, we gotta have- Bragging is important. Yeah, we gotta have our hype team with us because, man, 'cause it will break, and you're not gonna be hyping that. Yes. But I love it. So each one of those pieces, you, you, you make sure it's working, you hype that up, and then you keep at it. You don't have to, what, boil the ocean or... Okay, I'll stop with the analogy. Exactly. Like, you can take it a step at a time, and like I've... Since I'm very in the weeds on automation, I feel like we haven't done nearly enough analogies- ... this episode, so I'm gonna throw one more at you, which is around identifying good candidates for automation. Ooh. And the way I think of it is like- Okay, you have a park, and it's got all these paved little paths through it, cool. But everybody keeps cutting across the grass in the same spot, 'cause it's a lot faster or more convenient or in the shade or who knows. There's actually a word for this. It's called a desire path, and it's like those foot paths that form in the grass because enough people walk that way. So finding candidates for automation in your company are about finding those desire paths. What are people already doing as a shortcut, and automate that. I'm stealing that. Well, and I'm thinking about the dogs that run in their circle around the... Okay, I'll stop. But like it can help identify good areas where automation will help, like make the most impact without hopefully breaking anything. Oh, see, I said at the top, when I was a sysadmin, it was always just like, "Oh, we're doing the same thing again. All right." Mm-hmm. "Let's automate that." Yeah. But that's exactly it. But it's a more positive s- take on it, the desire path. I love that. I hadn't heard that before. I think it's the difference between if you're creating automation or scripts for yourself versus- Yeah trying to build it for a team or a department or a company. Ah, true. Yeah. 'Cause you have to find kind of the paths of least resistance because that's how adoption actually happens. Oh, yeah. That's a great point, and I do. I've been trying this whole season to s- stay positive and say really funny, like upbeat kinda things because, you know, again, we see a lot of fires. We see a lot of, "Oh, I have to do it again." But no, no, we're gonna do this because it's positive. We, we want to. Yeah. It will help us. It will save us time. Now instead of saying, you know, "Oh, dang, and I gotta do this again," you can be like, "Ooh, this could be automated." Nice. Then I never have to do it again. I love it. We talked a lot today, and I feel like I talked a lot today, so. Good. I, I think we can kinda recap what we talked about and then maybe kinda what we talked about in the season, too, before... Like I'm not quite ready to let you go yet. I'm not ready to go. I know. This was so fun. And we, we ma- we made up some questions ahead of time 'cause we do have to focus every once in a while, darn it. Yeah, a little structure, you know. Yeah, a little structure. I mean, I rage against it, but whatever. Here we are. But I, I have to s- say right at the top, though, it is, it has been absolutely amazing. This is the first time I've ever done a podcast. I was interviewed a few times in the past, but this is the first. So watching the behind the scenes, learning how these things get created, oh, you guys, that was such a wonderful thing for me. It's so fun. But I just wanna make a little personal note real quick. I don't think people talk about that enough, especially if you're brand new. I know. So if there's anybody listening, if you're ever invited, do it. Do it. Oh my gosh, do it. It's so fun. Um- It's so fun. We don't bite. Right? I mean, it's great. This is so great. Um, yeah, I mean, I can't speak for any other podcast people. But these people, oh my goodness, they're amazing. And they kinda like... You know, you guys are so, um... You brought it out of us. You know, even you, Emily. We, we, we... All this information in our heads that we do every d- day that we don't realize that other people might care about might like to hear about. So on that note, I can't help myself but be a little bit biased myself since- Mm-hmm ... virtualization is in my heart, and has been in my heart for a really long time. So networking, edge. Oh, okay, I have a cheat sheet in front of me of all of the episodes. I'm not gonna read them all. I love all of them. I can't pick one. Yeah. It's like picking your favorite cat. Wait, I don't have that many cats. Um, but I really love the virtualization one, because it explained a bit more about where OpenStack and OpenShift fit if somebody is rolling out AI workloads. And I know we gotta stop talking about AI, but it's what everybody's talking about right now, so whatever. I mean, you could put any application, but I'm just gonna say AI is the application. Where that fits in the current, like, technology stack that we have, right? So that was, to me, I'm a little bit biased toward my friend Maria. What was yours? Well, I'm obviously super biased, so automation is going to be my favorite topic. So I'll, I'll talk a little bit more about, like, what I learned the most about. Okay. Which I think I would say containers. Because I knew, like- How to use them, what they were, what they do in, like, our processes and things. But like, I, I never really talked much about the history, how they came about, the ins and outs between them versus virtualization. I thought that was all really interesting to talk about. Awesome. Yeah, and, um, we did say the word pod a couple hundred times on that one episode, right? Okay, good. Okay. Hopefully some of- I was counting. I know. I, I totally missed it on one of them, and I couldn't believe it, so there's a whole bunch of things. The technology is a lot deeper than, you know, a half an hour podcast. Oh, I love it. Uh, okay, so what was something unexpected that you heard during the season? Oh, I think probably any analogy we ever use probably qualifies. I think me saying spaceships that go underwater was one of the most unexpected. That and the thing where I said, um, uh, keep your UDEV environment separate from your production. It was UAT, y'all. UAT, not UDEV. We're not talking about dev- Like, I mean, in the ver- If you stretch it as much as you possibly can, yeah, you don't want your USB drive that you plugged into your machine- on production, but that's not what I meant. I meant UAT. So yes, our analogies and the crazy stuff that came out of, came out of our... I love some of the phrases. Like, and you go back and you listen. Oh, and it's always a trip to, like, listen to it after the fact 'cause it's like- Oh ... "Oh, that's not what I meant." And that's one of the reasons that I'm like, guys, you... I- if anybody's out there, like, "I wonder if I should do a pod-", yes, you sh- you learn so much about yourself also, like what you think you said versus what you actually said, but it's okay. Like, it all fits in, and then, you know. It's so true. Again, the folks that do the editing are amazing, so I'm gonna keep throwing flowers at them. You know, to be serious, you know, topic-wise, I think the one most surprising was probably, like, edge devices 'cause, like, you know- Yes ... that they're in everything and they're everywhere, but, like, the true depth of, like, what an edge device can be and be for- Yes ... was a little, I think, a little surprising. You don't think about it And then the other thing with every single topic that I'm looking at is the definitions change. Mm-hmm. So when we walked into that edge, edge, when I was thinking about how I am right now working on little tiny devices that do a lot more of the computing closer to each individual, whoever's interacting and sharing data and whatever. But 10, 15 years ago, it was just taking the data and sending it somewhere else for it- Yeah to be processed. And that, we got through that. I think we did a pretty darn good job explaining that during the edge computing episode. Yeah. But I, I wanted to highlight again how important, and, like, even though I work in this and I do this all the time, I guess that was something unexpected for me, was how that has changed, right? How that's changed so much over the years, and you don't realize. Like, it has changed, and I think, like, there's nuance to what that change means- Yeah ... like, from a, a technical aspect, and I just think it's really fascinating to learn how that works. Right. And nobody ever gets deep enough into any of the networking. Thank you to our networking sysadmins and DevOps and folks that are listening right now. We have much respect. Oh, yes. Thank, thank you for your service. Everything in there is always new to me, even though I've also been doing that for 20 years. It's all new every time. So. True facts. All right. So I think- And- Sorry, go ahead. Oh, no, I was gonna say, do you wanna wrap up? Because I was... Well, it was do you wanna learn anything new? But this episode I learned quite a bit new from you about automation. So. Oh, yeah. I learned a ton about, like, operating systems and images and things. Like, I take them for granted 'cause I just use it. Yeah, I mean, I'm, I'm trying to figure out how to, um, sum up this one that we just did because we had the r- script flipped a little bit. Most of the time I feel like you're doing an outstanding job. Well, I, I don't feel it. You are doing an- ... outstanding job summarizing all these incredibly complicated things. This episode, I was trying to take in everything that you were saying. Yeah, yeah, yeah. So now I feel like I'm- I'll let you take a shot. Well, I'm trying, but it's... There's just so much. So we'll, we'll go back and we'll say again that compliance is not about checking boxes manually, of course. Mm-hmm. And this, it's about scaling identical, predictable environments. As much as we are not identical, predictable humans, we need to- ... to have scale predictable environments. Automation is a great way to accomplish that. We know what your favorite product and my favorite product is, so I won't say it. Of course. Um, we also are now interacting in this world where folks are rolling out, um, whatever the definition of AI is today- Mm-hmm um, all over the place. And so these tools are so good for building those guardrails around that automation, keeping the systems resilient, consistent, and cost-effective. Uh, how, how am I doing? Exactly. No, you're nailing it. And I think, you know, automation doesn't just get rid of manual tasks, it can standardize your processes, which go hand-in-hand with those identical and predictable environments, and that you can account for failure with rollbacks built straight into the architecture that way. Yes, I forgot to say the rollbacks. Yes. Very, very, very, very important. Yeah, you only have to do it without once, and then you won't forget again. Yes. And I think the last thing I would bring up would be that you don't automate just for automation's sake. Oh, yeah. You should have a strategy around it because that's what keeps it from conflicting with itself. Like, all of your automation will function a lot more smoothly and effectively if it comes from a plan. I love your summaries. Could listen to them all day. I did. Well, you did fantastic on it. Well, thank you. So I, I appreciate that you were listening to me ramble on this one, for sure. It's amazing. Emily, this has been so fun. I just loved every minute of this. It was so fun getting into some technology that I haven't gotten too deep into recently, or never in some s- cases. And then also just to be able to tell some of my stories. Thank you all very much, and I would like to personally thank everybody that listened. I love, if you can't tell, I love telling these stories. I will tell them any time. I am my father's daughter. My dad is the same way. So shout-out to my dad for impo- uh, you know, um, sh- passing that down to me. So I will happily tell all of my stories, and I will tell them again, and the fish sometimes gets bigger, and the fish sometimes gets smaller when I tell my stories. Um, but I have so many from over the years, and I'm so happy when anybody even pretends to care about them. Oh, no. So thank you so much. Emily, thank you so much. Well, and Jennifer, thank you so much for sharing them. This has been a super fun season with a lot of topics that I knew a lot less about. Uh, and I, I feel like I've learned so much, and I really appreciate you coming on, and I, I hope you come back around again. All right. Thank you, lady. And I would say the same to all of you. You've heard us talk. Now hit us up on social media at Red Hat and use the hashtag #compilerpodcast. And that does it for this episode of Compiler. This episode was written by Kim Huang. Thank you to our guest, Chad Ferman. Compiler is produced by the team at Red Hat with technical support from Molly Brock. If you liked today's episode, please follow the show and leave your review on your platform of choice. Until next time. Bye everybody.

More about automation and management

Automation reduces or eliminates repetitive tasks, typically in industries such as manufacturing and automotives. 

Blog post

深入探討政策即代碼(Policy as Code)：為何是現在以及如何提供幫助？

2024 年 4 月 18 日

About the show

Compiler

Do you want to stay on top of tech, but find you’re short on time? Compiler presents perspectives, topics, and insights from the industry—free from jargon and judgment. We want to discover where technology is headed beyond the headlines, and create a place for new IT professionals to learn, grow, and thrive. If you are enjoying the show, let us know, and use #CompilerPodcast to share our episodes.