Reflecting on a Year of Cloud Networking with Cables2Clouds - C2C029

Chris Miles: 0:00

I think we got some really great responses. Some of them were super technical in nature and kind of expanded, and then some were completely out of left field. I didn't expect to get some of these answers at all. Welcome to the Cables to Clouds podcast. The option is on the rise and many network infrastructure professionals are being asked to adopt a hybrid approach as individuals who have already started this journey. We would like to empower those professionals with the tools and the knowledge to bridge the gap. Hey, everyone, and welcome back to another episode of the Cables to Clouds podcast. My name is Chris at BGP, main on Twitter and only Twitter. And then we have Alex Perkins, who is at bumps in the wire on Twitter.

Alex Perkins: 0:57

He's not on X.

Chris Miles: 0:57

He's never joined X, he's only been on Twitter his entire life. Tim McConaughey at Juan Gómez on Twitter as well. So today we have a very special episode that we put together for you guys. So we're recording this in March 2024 and we have literally just I think the other day celebrated one year of this podcast.

Chris Miles: 1:27

So we've included all the sound effects for the noise makers and party hats and things like that. So, yeah, so we've officially hit one year and it's been amazing. So we started. This came in under A1, who's been so supportive, and we've been reaching out to listeners and the responses from you guys have just been so great. So first of all, a big thank you to you guys for helping us to get to one year.

Chris Miles: 1:52

But to kind of celebrate this, we wanted to do something special for one year episodes. So what we've done is, you know, we've had a lot of great guests on the show even within just the first year. We've really made an impact and resonated with our audience and things like that. So we wanted to bring them back and kind of do something fun. So for this episode we've, over the course of the past month, we've gotten together with a lot of our favorite guests that we've had on the show. We recorded a brief segment with each guest that we brought back on. So you'll see there's a whole list of them. We'll put it in the show notes, but we all wanted to ask them the exact same questions and kind of get their responses on a few different topics. So you know, we will get to that in a minute, but before we get into that, let's talk about our listener survey that we did earlier this year. So we don't want to spend too much time on this. We know you know yet another survey. I'm sure you know everyone has survey fatigue at times. So we do appreciate all the people that did fill out our listener survey. But we just want to kind of touch on some of the high points that we extracted from this to let you know that we are listening and we do want to incorporate some of that feedback into what we're doing today.

Chris Miles: 3:17

So first piece of the listener survey responses that we'll touch on is kind of the demographic data about you know who you guys are, who's listening to the show and etc.

Chris Miles: 3:28

Right, so one quick thing I'll call out, because this one's an easy thing and I think we put out the listener survey before we kind of changed anything with like adding the news segment and things like that is, we had responses from people about how long you typically listen to the show. Probably most of the responses close to you know, 45% are right in that 30 to 60 minute range, some people in the 15 to 30 minute range and other people saying it depends, which is good for us, because now we've started a new segment that comes out on a fortnightly basis as well, and that's typically. We try to keep that one anywhere from, you know, 20 to 30 minutes, not really going over the 30 minute mark. And then when we have a guest episode has who has something to showcase, we try to give them at least 45 to 60 minutes. So, you know, I think that's that's good information, where we're definitely trying to align to that. How about anything else you guys want to touch on with regard to the listener, demographic data?

Alex Perkins: 4:28

Yeah. So actually, like you know, looking at a couple of them, something like the age stands out right, like 50% of the listeners are 35 to 44, which I think all of us are too right, like we're all, we all fill that age range.

Chris Miles: 4:43

Tim's barely in there.

Alex Perkins: 4:45

I'm squeaking, oh man. Well, tim, you're in luck, because the next highest percentage, at 35.7%, is 45 to 54.

Tim McConnaughy: 4:57

There you go.

Alex Perkins: 4:59

And then we got 14.3%. The smallest chunk is 26 to 34.

Chris Miles: 5:04

So we need, we need. We need to touch in with the younger folks, the people getting in into the field. Yeah, if you know someone that's just getting into tech, just getting into cloud, specifically the recommend the podcast if they want to get into it, and we know we'd like to kind of expand out that direction.

Tim McConnaughy: 5:22

Yep, that's why we joined TikTok, yep, before I got banned. So before I got banned, yes, my band soon, yeah, no, pretty soon, as of by the time this comes out, maybe we'll be, who knows? Okay, so let's, let's do a little bit more about like, okay, so those is the age ranges let's talk about, like the position and kind of you know where they are in their career. The overwhelming majority of people who listen to the podcast are individual contributors. Not really surprised this is networking focused podcast. We tend to be very skew, very heavily on the individual contributor side. You know there's some upper management, director level or exact, even have a few apparently managers and, like I said, director level people. So yeah, it's great to have all of you and just you know, right on the back of that. Where are people in their career? Even split at 50% is seasoned vets. This skews exactly right with our with the response rate on the ages, right, so that none of this is surprising.

Alex Perkins: 6:26

So for sure.

Tim McConnaughy: 6:27

Yeah, exactly, and then you know it's either I've been doing this forever or I'm a seasoned vet, which I'm not sure what the difference is, but that's the overwhelming majority there.

Alex Perkins: 6:37

I will say let's get into that, you know. Vp president C C Sweet, let's start getting some of those listeners in here. Maybe our hot news takes will start bringing in some of those people some angry, angry people perhaps, maybe.

Chris Miles: 6:52

I think the episode we did with Joe would probably resonate much better with that audience, and then anything else we've done so far. But yeah, so that's good. So that's kind of the demographic data about you guys. So let's get into the tough stuff. What's your least favorite part about the show? I think one of the main things called out here is the. They just put quote how you doing, jim? Which, referring to you know the banter that we, you know, had at the beginning of every episode. It's like hey, tim, you know I've never talked to you, I haven't talked to you all week. How are you doing, man, when that's total bullshit? You know we've been.

Alex Perkins: 7:33

We've been in episodes about a trip to.

Chris Miles: 7:35

Japan. I did finally go. You went to Japan, he did it, thankfully. We've kind of cut that out. We don't want to kind of, you know, waste time with with a bit of that. But we've also had our news segment come in where we can kind of make sure we're maintaining the personal stuff and things like that. But we don't want to kind of feed that into our guest episodes. We want to make that all about the guests. You know they have something to say and we feel that's valuable, so we've kind of cut that out. Another piece that someone called out was long intros. And yeah, we get that. I myself was already before I think I saw this. I was like, man, intro is too long, man, and even it's only 30 seconds, but still too long.

Chris Miles: 8:19

So, yeah, too long we're going to release. We're going to release a new intro where it's the intro playing and the bottom is like that subway surfers game where the guys jump on the trains.

Tim McConnaughy: 8:29

I don't know if you guys know what that is, but this is a real game on it.

Chris Miles: 8:35

But yeah, so we've re-recorded the intro. Actually, we're going to very soon we'll have one that's that's much more condensed and concise, so we're going to definitely address that. One other thing is someone called out the language and saying that. You know, from time to time it kills the mood or dilutes our authority as professionals, et cetera. And while I will say that I understand, I understand the sentiment that that it's not necessarily the choice of everyone, that they'd want to speak like that or listen to people that speak like that, and I totally respect that and understand your opinion.

Chris Miles: 9:14

But, leading into kind of what people have said about their, their favorite parts about the show is the. You know, the personalities are canned or with each other are rapport, and you know I'll be completely frank with you guys. We started this podcast because we wanted to be able to sit around and talk with our friends about technology and things that we like, right, and if we want this thing to feel authentic, we need to talk authentically and talk comfortably. You know me, tim and Alex, I've been friends for a long time and you know I'm sorry, man, this is, this is how we're going to talk and I'm trying to be, you know, I guess, respectful with it and where it's not, you know, being added in for for no effect or anything. But you know, fact of the matter is that's just how we talk. So I don't know how you guys feel. You want to kind of touch on any other favorite elements that our listeners pointed out from the show, favorite or least favorite. I think we covered all the least favorite.

Tim McConnaughy: 10:08

Let's go in the favorite.

Chris Miles: 10:09

Oh right.

Tim McConnaughy: 10:09

Gotcha.

Alex Perkins: 10:11

Real, real quick. On the language thing, I just want to add something that basically you touched on, chris, but as far as like being, you know, sincere and having like a candid conversation, we've had a bunch of guests that I've just realized this as you were talking. We've also had guests that like during the middle of the episode it just comes up and they're like, hey, are we allowed to cuss on here, right? So that's, I think that leads itself to show that, like sometimes it's just how the conversation flows when you're just in the game.

Alex Perkins: 10:40

Yeah, exactly, there's an authenticity factor.

Tim McConnaughy: 10:42

You definitely lose when you start. I won't say censoring I think that's a strong word but when you kind of ask people to self-regulate the message that they're trying to give For sure.

Tim McConnaughy: 10:53

So, yeah, totally, and I get what that. I get why you probably can't listen to this. You know, in the car with your kids certain episodes some some time. Certainly All right. So yeah, let's jump to the what do you? You know what people report as the thing they liked most about our show. The very multiple people said you know the host, candor, personalities, chemistry. We really I speaking for myself I really appreciate that. That's obviously a huge benefit or bonus to why, to knowing that you know we're kind of hitting the right mark just by being authentic. So that's a big one for me.

Chris Miles: 11:28

They must not be listening to the episodes where Tim talks.

Tim McConnaughy: 11:31

Probably not. I usually stay pretty quiet on these, so you can only find maybe one or two, yeah.

Alex Perkins: 11:38

Yeah, I think that's the best part. I think, honestly, since we've broken out the news episodes, it really comes out a lot more in those, because we're just riffing right off of the crazy news stories or you know, hype cycles that are everywhere. So I think, if you're, if you like the personalities, definitely the new show brings out a lot more of that sometimes.

Tim McConnaughy: 11:59

Yeah, and I, for one, I'm glad that we broke it out, because we definitely had feedback that you know, hey, this time stamps your episode and people who downloaded a year from now are going to suffer through the news. So we completely agreed with that and we broke it out. Honestly, it's performed really well. Just let you know, like not really did the survey, but since we did that, it's actually been performing extremely well. So thank everyone, both for continuing support us on that and also for just, you know, giving us that feedback, because we definitely want to re-incorporate, we want to incorporate your best feedback where we're ever possible. You wanna grab one of these other ones, alex?

Alex Perkins: 12:35

yeah, other favorite parts, right, obviously, the topics that we cover. This is super beneficial for us, right, so we can plan future shows. Overwhelmingly is pretty much all the same which guests, technical episodes and topical episodes. And I think, right, we used to do quite a few round table episodes and I think we kind of use some of this feedback to really Start targeting much more guest episodes and you've talked about this a bunch but, like system designs, you know, like much more episodes focused around specific topical things.

Alex Perkins: 13:10

So there's much more coming that cover the fall into these three.

Chris Miles: 13:14

Yeah, I think you know the cat.

Chris Miles: 13:16

The last year, I think we've had some stuff really perform really well and obviously really resonated with the audience that are that are more topical than technical, which is good.

Chris Miles: 13:27

Like, I think, the episode we have a Joe on this on, I think, was obviously not not overly technical, but it really kind of touched on something that is adjacent to everything that we do within within technology, right. But then we've also had one where we kind of just really dive into you know, a technical topic, and that's also done very well. So we kind of want to maintain that balance where we're we're touching on things that are, you know, may not be technical in nature but still very important to our day to day you know lives, professionally and personally. But then we also will have some very you know solutions focused stuff, right, because that, I mean, that's that's what pays the bills for us, right. So we want to get back to that and, tim, I don't know if there's anything you want to add on that. I know that was kind of something you were spearheading. Is anything you want to touch on there?

Tim McConnaughy: 14:17

Yeah, I mean, like the biggest thing I want to say is that there's we have a pretty good hopper of an unreleased technical content with more on the way, so On the way, so I couldn't say I have said it better. We want to, you know, on right, right in the middle, as much as possible, we want things to be topical and accessible to our whole audience, which may not be Super technical on in the cloud right now, but we also want to actually offer that technical knowledge and content for the people who are already doing cloud networking. So we're trying to strike a balance and we're very we want to know basically are we hitting the mark from from our listeners?

Chris Miles: 14:53

Yeah, just something I want to add to that real quick. There was also a response in the survey that you know we commonly will use, you know, acronyms and things like that and not explain what things are. Yeah, that's on it, that's on us, and if you know, we've we've set up in the beginning we were a podcast, you know, help, you know, focused on hybrid cloud networking, multi cloud networking, and a lot of people are very early in that journey and may not know anything and everything how we? We certainly don't know everything, so we need to be better about, you know, actually articulating Things that we're talking about and not leaving people you know like you know, I'm knowing what we're talking about, so we'll definitely be working on that.

Chris Miles: 15:36

Anything else, we want to touch on the listeners survey real quick before we move on. I think that's all of it. I think one other thing that that stood out to me is we had a question in there about where people are, where they are at work, necessarily like what kind of projects and things like they're working on, and 50% of exactly 50%, so that you guys are migrating to a cloud provider. I wonder if that means the other 50% are already migrated to a cloud provider.

Chris Miles: 16:03

I don't know, maybe totally maybe they haven't, but that thought that was an interesting point. So it was curious to hear from you guys. Right, if any of this kind of sparks anything that you want, you know how, if you want to talk about it, if you want to, you know we can talk to you on social media about it, we can talk on the podcast. You want to come on. You have a topic you want to discuss with us? We would be happy to have you, so let us know. Or if you just have ideas, definitely always open to that, always have an inbox that's ready and waiting, so just let us know.

Chris Miles: 16:39

Alright, so with that, let's, let's get to the fun part, the episode that I just told you about at the beginning. So we have our elite group of guests coming back on to answer some of our A quick set of questions and you know, I think we got some really great responses. Some of them were, you know, some of them were like super technical in nature and kind of like Expand it, and some were completely out of left field. I didn't expect to get some of these answers at all, so it was a very good mix, so Eager for you guys to hear it. So without that let's. Let's get into the episode. Alright, first question let's get a very quick who are you and what do you do?

Eyvonne Sharp: 17:25

Hi, I'm a bond sharp and I lead the north american infrastructure modernization team for Google cloud.

Pete Lumbis: 17:32

I'm Pete Lumbos and I'm Docs engineer at upbound my name is William Collins.

Will Collins: 17:37

I work at a network startup called Alcira where I help out Doing anything from testing and validating automation things you know, think like terraform, you know stuff like that and I also ramble a lot at lots of events and I prototype lots of things around cloud networking. I've even been known to help out on the sales side every now and then. Other than that, I run the cloud gambit podcast and I do a little bit of LinkedIn learning content, which is a blast, which I'm not going to be able to do as much anymore Because we're actually expecting our third child. So life is going to get much more busy congratulations.

Chris Miles: 18:20

Thank you, congrats bro.

Will Collins: 18:22

No, no way like a podcast to announce stuff like that. Eventually we'll tell family, so Find out from the internet.

Steve McNutt: 18:31

So my name is Stephen McNutt and I've been like a computer nerd for a very long time. Like my first job for people pay me money was Microsoft and that was Windows 95. So my first, my first IT job you could install Windows from a floppy disk at. Still, that's how I'm like a gray beard, I've been around for forever. So currently I work for Cisco Systems as a cyber security technical solutions architect.

Du'An Lightfoot: 18:54

Hey, what's up? My name is Dwayne Lightfoot. I'm a developer advocate at AWS.

Shala @GiftedLane: 18:59

I am Shala across social media. I'm gifted Lane. These days, more and more I see myself as a egg you tanner. People like to call me an educator. I like to tell them no, I'm not. It's like sharing my journey. But apparently I'm teaching people, but in a fun way. So fine, I'm an entertainer. I like to also think of myself as a cloud advocate.

Andrew Brown: 19:23

These days, Everyone, I'm Andrew Brown and I create cloud certification content, run cloud boot camps and just overall cloud cloud drama online.

Chris Williams: 19:33

Everybody. My name is Chris Williams, I am the developer advocate manager for Hashi Corp for North America, and I am also an AWS hero and a VMware v expert.

Alex Perkins: 19:42

Alright second question for you what is something that you feel worked better on prem than it does in the cloud?

Eyvonne Sharp: 19:48

Well worked better like that's an interesting question so that there are some use cases that only makes sense very, very close to the user. So it's why we still hear a lot about edge and even your cloud providers have edge solutions. So there are instances where you have a very latency sensitive application, where you have an app that or there's a business risk if there's a loss in connectivity. There are very their locations that require low connected or that have low connectivity and there you have to have a solution that can be on premises. It's just a requirement. So I would say, when I'm thinking about what maybe doesn't make sense to run in the cloud, those are the first, the first use cases that I think of.

Alex Perkins: 20:38

Yeah, that's a great point. Especially, edge is becoming a topic that I'm seeing people talk about more and more like every day, it seems like, so I think that's a good one.

Eyvonne Sharp: 20:46

Yeah, it's huge in retail. If you have, you know, think of a subway. Right, there's a subway in every little corner of the country and a lot of those corners of the country don't have great connectivity so you still have to be able to process orders even if you're having connectivity challenges and so a lot of edges, huge in retail. They're also big use cases, even in telco or manufacturing. But yeah, there are lots of times, in circumstances where you just have to have your compute resources really, really close to the end user.

Chris Miles: 21:25

Yeah, manufacturing was the first use case I thought of, because that's a lot of that yeah, depending on the, how big the or how you know grand the, the OT network is, some of that stuff could be, you know, life or death type stuff, like if those systems can communicate we're gonna get hurt at the workplace, type thing. So yeah, definitely make sense.

Pete Lumbis: 21:43

I'm gonna have to go with cable management on that one.

Chris Miles: 21:45

I don't know though, because, like, if you go in those CSP data centers, those, those, that cables, those cables are run pretty well.

Pete Lumbis: 21:52

Yeah, I mean your virtual. Your virtual cable management is trash.

Tim McConnaughy: 21:55

Yeah, that's true. I can't, I really can't argue with Well hold on virtual cable dumpster fire like nine different colors. Total rats now yes yeah, we just rip it out when we don't want it anymore. It's, it's pretty bad. We just clip the cables with big scissors whenever we don't want them anymore.

Pete Lumbis: 22:13

Yes, no work for cloud company. There is a sandbox account that's burning like forty thousand dollars a month right now. I can guarantee.

Tim McConnaughy: 22:21

We, yeah, we distribute ours amongst the entire organizations to get the forty thousand. So yes, absolutely true.

Will Collins: 22:31

Well, there, you know, there's a ton of ways and a ton of different things that come to mind, but if I had to pick one, I'm not gonna use that one. I'll say routing, routing, routing is a good one. So this is relative. You know, there's a lot of external factors here, but with, like, on premises networking and I'll preface this by saying, you know, this is might not always be a good thing, but you are in full control over your routing topology. How many routes do you want to advertise? What do you want the next top to be? Because on premises, you can get creative with traffic engineering. Okay, you know, routing traffic from point A to point B, how do you want to do that? And cloud, not so much. You're limited to the system in which the cloud is built for you. A lot of flexibility doesn't exist, and I guess it really comes down to control. Yeah, control over convenience or vice versa. Again, completely relative.

Alex Perkins: 23:28

I like that there's no quotas on-prem right.

Will Collins: 23:31

Not really other than the size of the box that you bought. You know how many routes. You know everything in cloud. It's like okay, route table, like all the different things have some limit that you're working within the confines of. Yeah.

Chris Miles: 23:44

I don't know any limits, but you'd be it's. You know it's funny that we talk about you're like oh, in the cloud. It's great, because the only routing protocol that's present is BGP, but it's also something that's not present in many places where there should be regular networking right.

Will Collins: 24:00

Yeah, I remember transitive routing when I first, the first time I was ever trying to help out a cloud this was before we even had cloud teams, like long time ago, and I could not wrap my head around it. Like I'm in this VPC, I got these subnets I want to route and I want this other VPC to be my next top to get to this other thing. Oh, can't do that. You know something you take for granted and the data center not existing in cloud unless you insert something else.

Chris Miles: 24:27

Yeah, I mean. That being said, did you feel like routing is? Is it actually a problem? Or is it us like the engineers at heart? For us like just the nerd? Yeah, we don't have the nerd knobs. We want to be able to control this stuff when we can't. Is it a good problem or is it a bad problem? I guess I should say that.

Will Collins: 24:46

I think it's a bad problem. I think it's gotten a lot better now with the likes of you know where you, you know you work, chris, and where I work, there's a lot of good startups that are innovating in the space. But thinking back, like the technical debt over the years, like site to site VPNs were the first thing I ever did and it was just the site to site VPN from some box on prim to a VPC. You know multiple AZs, you know two tunnels, and then it was like the transit VPC design, which was maybe the worst design ever conceived. And then transit gateway came along and made life a little easier. Well, a lot easier, you know. So, yeah, there's been a progression there.

Will Collins: 25:21

The cloud providers are trying to help, but it's definitely a problem because it pigeon holds you into these, these hub, and spoke topologies everywhere, like per region, per cloud. You've got all these hubs and since you can, only you know you're basically doing VPC or VNet, peering back to the, the hub VPC, where all your hybrid kit is to get you back out to you know somewhere outside of that cloud. And yeah, it's just a design pattern. That isn't. I don't think it's necessarily a good thing in all cases. But yeah, it's a trade off, of course.

Steve McNutt: 25:54

Network access control. So what's cool about, okay, the cloud, what's good about it is like there's no layer two, so you don't have, like any of that kind of nonsense to deal with. But also you don't, like you don't control things, so you have to. So if you want a capability, it has to be packaged as a service by the provider, right. But like if you control the actual network, that means like when, at the when a device, at the point of attachment, you can bring whatever fancy tooling you want to that right, so you can say, oh yeah, I want you to have a certificate and I want your computer to have a certificate and you'd have this software and we're going to make you do MFA and oh yeah, I want you to stare into this. Like you know, iris scanner. The point is you have complete freedom over what you want to pile on, as far as you know, deciding what's going to get, be able to get access and pass packets.

Du'An Lightfoot: 26:46

You put me on the spot. What is something that worked better on Prune than it does in the cloud?

Shala @GiftedLane: 26:52

You know that's a good question and I have to go to because of my, my work experience, I'll have to say if it's something that is like HIPAA or regulated data like that, you need to bring that on print.

Chris Miles: 27:09

Yeah, I feel like a lot of those regulated data pieces. It was kind of easier to. Yeah, you can you always enforce that. You always have the option to just rip a cable out or something if you really need to, and you know fire safe situation type things. So that makes sense.

Andrew Brown: 27:24

I mean, I think right now, if you're getting into generative AI, you want to run your own models. It is super painful and super hard to run these things on SageMaker, on vertex AI, on Azure, ml, machine learning, whatever, whatever. And I find that if you have a good graphics card like I just bought an RTX 3080, and you know things just run, it's less painful and I've just not bought into this whole running your ML models or, you know, playing around with this stuff in the cloud.

Chris Miles: 27:56

How long do you see that lasting? Do you think that's going to be something that becomes usable soon, or do you feel it's hopeless?

Andrew Brown: 28:04

I don't know, like at the end of the day, you're going to have to run something, but whether you are actually going to use their tooling or use something cloud agnostic what's the thing called that. It lets you, it's like pod something that allows you to pod runner, pod something while you're on your ML models. I feel like that's a lot easier because you can take your local environment and better replicate it in a cloud agnostic way. But yeah, I'm just not liking the tooling in general for the online ones and I don't have good hope for them because they keep changing it.

Alex Perkins: 28:34

Well, there's also Nvidia just announced like a way to run your like on your laptop, right, and I think we're going to start seeing more and more of that. Oh, the.

Andrew Brown: 28:43

RTX check thing Coming up very quick. I just downloaded it yesterday.

Chris Williams: 28:46

Oh, so low latency applications worked much better on prem than in the cloud things that I mean. To go back to the monolithic application paradigm, things that were tightly coupled and if one thing broke down, if everything was in one place, then you could quickly find out what that thing was and what the breakpoint was. It's much harder to do resilience type monitoring for monoliths up in the cloud. I might have worked for a military company, whose name shall not be mentioned, that if their on-prem application was greater than 10 milliseconds then the nuclear launch codes became corrupted and that was a bad thing.

Tim McConnaughy: 29:31

All right, here comes the third question what is your cloud hot take?

Eyvonne Sharp: 29:35

All right, you guys ready for this one? Not every application needs to be modernized.

Alex Perkins: 29:41

Oh my gosh, she went there.

Eyvonne Sharp: 29:43

I did go there. So there are a lot of instances where you have an app, it works perfectly fine, the way that it's written, and maybe the platform needs to change. You can absolutely run that app in the cloud. But you don't necessarily need to completely rewrite the app from the bottom up, make it microservices, change your database, all of those things. I think one of the things that we as technologists sometimes downplay is we don't always think about the business side of that, the cost, and a lot of times it's an opportunity cost.

Eyvonne Sharp: 30:20

If you're going to rewrite an application, just is it worth it to the business? Are you going to? If it's an application that needs to change and grow over time with your business and it's very core to what you do and you need new capabilities, there's a very strong case for modernization there. But if it's an app that does the thing, it does the thing you need to do. Well, oftentimes those apps can move to the cloud and run NBMs, can sometimes be modernized, even into containers, and can even run in a serverless environment without modifying the application. That's something that we don't talk a lot about.

Alex Perkins: 31:04

I love it. Our mutual friend, will Collins, did a talk a while back where I just loved a slide he did. He basically had a picture of a bus and he was explaining the story like he told his wife oh, I got us a new family vehicle right, and it was like a school bus instead of a mini van or something. So it's just, it's very fitting for sometimes people over there.

Eyvonne Sharp: 31:26

Yep. Right tool for the job. Right tool for the job. What does the business need, what makes sense economically and how do we continue to keep everything working while we also grow, change and, to use the other magic word, transform?

Chris Miles: 31:42

Yeah, I think that's good because it's like, obviously there's a strong focus on things that we should do we should modernize apps but there's also you got to take it both ways right there's also things that you shouldn't do. If something's not part of your strategic roadmap, then you probably don't need to put in the effort now and waste time and resources. So, yeah, totally agree.

Pete Lumbis: 32:03

There's a lot of value to the cloud, but I think a lot more folks should be building data centers than they think they should. I think a lot of people are failing to realize that the cloud is basically pick one of bandwidth, compute or storage, and a lot of people are using all three and paying twice X or three X in OPEX against some KAPEX.

Tim McConnaughy: 32:26

Yeah, I think that's entirely valid actually.

Alex Perkins: 32:31

Do you think, as all these the AI talk right now, a lot of that's going on. Do you think that companies are going to actually start building their own on-prem stuff to run these AI workloads, or is that going to stay in the?

Pete Lumbis: 32:43

cloud. I think it will eventually, for two reasons. I think number one in the long run it will, but right now you can't buy a video card, so it doesn't matter.

Tim McConnaughy: 32:51

Good luck.

Pete Lumbis: 32:54

Only the cloud companies can get them, so they're the only ones that are building. But I think, number one, the cost is insane, the markup there. But number two, I think a lot of it comes down to privacy. The data that's building and training your model is your core IP. I kind of wish you washing that through somebody else's model.

Tim McConnaughy: 33:17

I wouldn't be surprised if the CSPs lean heavy into the idea of a walled garden model, though if they haven't already.

Pete Lumbis: 33:24

You hear your private sandbox of LL and whatever you could do that today you can take your own models and you can train them or refine them, but it's like anything else. Is it worth it? What's the value there? What's their turn on investment? I think hasn't totally happened yet. Is people realizing that that training is going to become crown jewels for the companies that get right? You're not going to want to put all that data on your competitor's platform or even a place where it could potentially leak out Again. This is some of that cloud fear-mongering that happened in the early days. You can't use an S3 bucket because what happens? There is security, but I think for a lot of the companies it'll be easier for them to build a cluster, run and train their own models and then protect that model locally.

Will Collins: 34:16

This is a real hard one. I got a lot of these. For the most shock factor, I'm just going to say artificial intelligence is massively overhyped. Ai is beyond the shadow of a doubt. It's definitely a game changer in a lot of areas. But networking, though an area of tech that you rely on to be deterministic. The investment in capital is going to be substantial. You're including the cost of all the resources, the training, the maintenance. We live in a world of cost and recouping ROI. I think the economic considerations are going to challenge the notion that AI is universally viable for networking, especially beyond the read-only and observability use cases. I think for read-only and observability, that's one thing, but I hear conversations sometimes yeah, AI is going to be running the network in two years it's going to be rerouting traffic and I don't know about that. Good luck.

Chris Miles: 35:24

Yeah, I definitely see the advancements at least, is how we operate. How we operate how we function in our day-to-day jobs. But the actual acting on that within the system, the network or whatever the compute virtualization system that seems so far-fetched it's only going to take one event for one C-suite executive to rip it right out, everything out.

Will Collins: 35:52

Yeah, 100%. And you got to think this stuff is only as good as the data you're giving it. You give it too much data, your results are skewed. If you give it too little data, you're not going to get any outcomes out of it that you need. You got to feed it the right amount of data. There's all sorts of things that we're still focused on in the network side of the house making sure internet's reliable and things like that, making sure DNS doesn't break. How long have we been trying to make sure that DNS doesn't break? Well, since I actually started working in this industry way back, same problem every day.

Chris Miles: 36:26

AI needs to fix DNS.

Steve McNutt: 36:27

That's what needs to happen for sure, I'm still really salty that logging is so expensive and I honestly think it's a situation where the federal government should probably step in and do something about it, because it creates an unnecessary security risk for customers.

Du'An Lightfoot: 36:44

My cloud hot take is that you have the ability to build whatever you desire because you've had access to compute, you have access to networking, you have access to storage, you have access to databases, you have access to serverless as well as generative AI. The possibilities are pretty much endless in the cloud.

Shala @GiftedLane: 37:03

Cloud is dying. No, I'm joking Cloud hot take. The first thing that comes to my mind off the cuff is I remember when network engineering going to get your CCNA and all that kind of stuff that was the hot thing to go do. Then it was cloud and now it's AI. That's my hot take. I think these things kind of unfortunately, I feel like come as fads here and there. I think people are moving on from cloud and now it's AI. One thing that's always funny to me is the whole. At the end of the day, even though we say it's in cloud, it's still in a data center running on machines.

Chris Miles: 37:44

Yeah, just out of your jurisdiction. I think, piggybacking on the first thing you said, I think cloud is also what advertises this thing. That's so easy and it's so not. It's so complicated. There's so many things that go into it. Yes, there are things that are easier, but at the same time, there's also things you didn't have to worry about because the landscape is not the same. There's obviously levels of inherent security.

Shala @GiftedLane: 38:12

You can call it built-in to things that you don't have to control. Yeah, exactly. The other thing piggybacking off of that is the fact that money a lot of people are moving off the cloud because it wasn't as inexpensive as they were made to believe. I feel like with physical equipment on-prem, I think it was easier to calculate costs.

Tim McConnaughy: 38:38

Yeah.

Shala @GiftedLane: 38:39

Cloud is a little bit hard and you kind of just get slapped in the face with a bill that goes O-shaped.

Tim McConnaughy: 38:46

For sure.

Chris Miles: 38:48

There's a reason FinOps is so big. There's an entire level of business about maintaining your cloud bill. I don't think that exists on-prem, because it's always predictable from the OEMs you're working with.

Alex Perkins: 39:01

It's cut and dry. Yeah, you just got a quote and there you go. There's no quote for I'm going to spin up. A dev wants to spin up something to test it out. You don't get a quote to you that you get approved and signed off on.

Shala @GiftedLane: 39:16

It gets spun up, Someone forgets to turn it off. They decided to try all the bells and whistles and maybe good policies weren't set in place to check that kind of thing. You get that outrageous bill and you're just like wait a minute.

Andrew Brown: 39:30

What is my cloud hot take? I think that I should probably do the thing that I'm most known for, which is about cloud certifications. I think you don't need a cloud certification. I would probably say cloud certifications are useless if you are trying to break into the cloud and we really need to do something about this.

Chris Williams: 39:47

I have so many spicy hot takes you got to pick one, I'm sorry. Give us the hottest one, give us the hottest, spiciest. The current paradigm for service creation in the cloud hyperscalers is not conducive to enterprise level systems. Creating creating multiple two pizza teams that each have a KPI of service and feature releases that oftentimes contradict each other Is a really shitty way to get enterprise adoption for people that just want things to work.

Chris Miles: 40:21

How often are you seeing that happen? Like is it? Is there a server every day?

Chris Williams: 40:26

Constantly. I mean, if you are a brand new startup and you're figuring things out and you've got a couple of accounts and and you're just trying to make things work and stand up, that's one thing. If you are, if you are an enterprise consumer, a large fortune 100 that has multiple thousands of microservices trying to work with each other and multiple software development teams that are trying to make things just go, and you have 200 plus services with different API interfaces, with cross functional services, that all will. Lambda does it this way, but Fargate does it that way, and and and. The reason that that is the case is because, internally, the way to get promoted at a large hyperscaler is feature releases and shipping, new features, the order, yeah, regardless of communicating with the other two pizza teams, yep. Then it just turns into a big fucking mess.

Chris Miles: 41:24

Alright, and let's round this out with the last question. What's the biggest hurdle you had to overcome, personally, when changing your mindset from how things are done on premises to how they're done in the club?

Eyvonne Sharp: 41:35

I think this one's twofold and it's it's more an argument that I've commonly heard. One is that it's always more expensive to run on someone else's infrastructure, and you hear that a lot like how can you run the same thing on somebody else's infrastructure and do it more Cheaply? You know economies of scale, all of that, but I think the other, the other side of that conversation is that things have to run differently in the cloud. You know, we, we well.

Eyvonne Sharp: 42:08

I'm of an age where I learned to do network security, network infrastructure, in a particular way.

Eyvonne Sharp: 42:15

You know you have a, you have a gateway firewall that's on the edge of your network and that provides all your security. You know there wasn't even a concept of a WAF or, you know, load balancer with an integrated security or DDoS. None of those things like existed back in the day, and I think what you have to begin to grasp is what the new components are and how you achieve the same capabilities with new technology as opposed to. You know, because the mistake I see a lot of folks make and a hurdle is you know well, I have to have a Seven firewall appliance on the boundary of every VPC, for example, and I have to do that because I'm re implementing the mental model that I have for on prem in the cloud and if you do that, it will be cheaper and it won't work well and you'll fail Ultimately.

Eyvonne Sharp: 43:17

So you have to embrace a whole new set of architectural principles and you have to rethink about the components that you need to make things work. It is not like let's just Virtualize this on premise function and put in the cloud. You have to embrace, really, a whole new set of primitives, that there's still some fundamentals that are going to apply no matter what. But you really have to understand what a cloud architecture looks like as opposed to an on premises architecture, and to reform how you think about what An architecture looks like.

Alex Perkins: 44:02

I might need you to come back on when we do a talk about some government guidelines around like tick designs. I agree, but, man, there's some mandates out there that make that very, very difficult to deal with.

Eyvonne Sharp: 44:15

Well, and that can be an interesting problem, right? Because a lot of times your, your auditors, are given a standard playbook for how something ought to look.

Eyvonne Sharp: 44:24

now you can Go and argue that although you don't have, you're not using the same components, that your architecture still meets the compliance standard because you're using, you know technology, why, instead of technology X the thing is Absolutely that's not an easy thing to do and unless you have an organization committed to and leadership that understands well enough to argue why that's necessary, then the quote unquote easy button and I put that in air quotes for a reason Is to just follow those old architectural patterns. But I put it put easy in air quotes because you're gonna do the hard work on the front end of the back right, like either you're gonna do the hard work to rearchitect or you're gonna do the hard work of Maintaining and supporting the suboptimal system over the long haul, and neither of those things are easy. One is more painful for longer.

Pete Lumbis: 45:26

Nuking it from orbit is a totally valid answer. In the cloud I think that was. You know, we come especially as a network engineer like coming from this like Plan it. Plan for the plan, review past eight people. Review the plan that you passed by eight people. Review the plan for the plan that you passed by people to review it and then still get it wrong. And then stay up all night fixing the VPN tunnel where you actually yeah still have a backup.

Pete Lumbis: 45:52

You flip the source, the destination to the VPN, right and now you're just like I don't know. I did it like six times in my laptop and it worked, and then we just blew it up when it didn't work and everything's fine. I mean, coming back to the. You know, I think more people should be building data centers. That is the downside. You know you have to have to allow you to have some of that mentality. But moving fast and breaking things Is really valuable when you can rebuild them even faster lean into the ephemeral nature holistically troubles you and communities is for losers.

Tim McConnaughy: 46:23

Is it working? It's got a blow to it.

Pete Lumbis: 46:26

Yeah, that's right. Just don't know you can't cluster every every hour on the hour.

Tim McConnaughy: 46:31

Shake off all the bugs.

Alex Perkins: 46:33

Some. What related? Pete, I remember when you were going through your learning Kubernetes journey and you're posting a lot of stuff and you're trying to upgrade a cluster at one point and it was like now I see why people use, use, manage clusters. It sucks.

Pete Lumbis: 46:50

And we had this engineering meeting and we were talking about, you know, our product is like this managed, it's managed crossplane, think of like managed Terraform, but different. And they were, like you know, trying to talk about the value of like the managed service, and like who here still runs a Kubernetes cluster Like, as opposed to just using our platform? You know everybody's hand goes up and like, ok, who's still? Things about the Kubernetes layer. You know bunch of hands are show up and I like who still spins up their own Kubernetes cluster? And I was like oh, just me, just me, like everybody else is running, like they're like I just run on the case yes, or a case yeah. And I was like, yeah, but I have a very fast, large computer, that it's. It's way faster for me to do that in vagrant. It just ate a whole bunch of shit to get to work the first time. You know, like halfway through Kubernetes the hard way, and I'm like I fucking hate certificates. This is stupid.

Will Collins: 47:40

That's a really good question For me. It was. I really I really think like looking back and I still actually still struggle with this sometimes is this whole like scale out rather than scale up approach to like your design and having, ultimately having like major self control and scoping things out? You know so, on premises, tech and culture, especially in the days of like it'll change management practices on steroids, you know that led technologists to get as much as they could at the start of a project or an initiative.

Will Collins: 48:14

In cloud, you kind of want to go the opposite direction and you want to. You want to start small and you want to scale horizontally. So instead of that one massive instance, you know you scale many smaller instances horizontally to meet demand and then, when that demands subsides, dial things back down. It's very much. It goes against everything that I ever learned from everybody in the data center days and we were building things, which was we need to get as much as we can, as big as we can, as fast as we can. You know that's the direction. This is a complete opposite way in cloud.

Alex Perkins: 48:52

I like that because it's also it's the same for everyone, right, like it's not just networking people, it's compute. It's like because compute people are right, notorious, so like, give me 24 CPUs for like this, this tiny little application that I'm building. So, yeah, I like that because it is. It is applicable everywhere.

Will Collins: 49:10

I used to do the same thing. When you know, going to the compute folks for like network appliance stuff, like, oh, we're, like I remember we were prototyping some like IPAM, like an open source IPAM system this was a long time ago and then you know different, okay, like a new syslog thing we want to do like syslog in G and like do filtering with syslogs before we pun them to something else, oh, I'm getting the biggest VMs I can possibly get for that. Oh, you know we're going to need the whole house. You know you might as well just give us that whole hypervisor, because we're going to need it.

Chris Miles: 49:42

Yeah, it's also difficult to because, like that, that's obviously something that's preached very hard whenever you move to cloud right, it's in CSP, unnamed as well architected, framework name, name, whoever it is they all say to do that right. But, like, if you're also wanting to leverage things like security from an ISV, like a you know a particular firewall vendor, they a lot of times they aren't there yet with their virtualization stack in order to support that right, like the time that takes you to stand up a new you know a scale out firewalls and bootstrap in and just push policy. The event's over. Like you didn't catch it. You didn't catch it at the right time.

Will Collins: 50:23

Yeah, that's a big trade-off you have to. You know, a lot of times it's either we're going to deploy the firewalls in the cloud or we're going to deploy them as close to the cloud as we can in a colo. So go to the big box approach and have that firewall actually, even maybe doing the multi-cloud routing, or having, like some other routers in the colo doing the multi-cloud routing, or hey, rolling with those virtuals and your hubs, all the hubs we talked about earlier, yeah, go that direction, favorite model.

Steve McNutt: 50:51

For me, like as a network engineer, I still struggle with the fact that there's no layer too right, just that, everything is. I just I can't help it. It's just like my brain just struggles to process that and I it's like it's like you just got to take like that whole everything below layer three and just lop it off to pretend like it doesn't exist. But it's, I still struggle with it sometimes.

Du'An Lightfoot: 51:13

Yeah, the biggest hurdle that I had overcome was logging into the console. When I'm working on routers and switches, I'm used to hopping into a router. I'm used to hopping into a switch doing a show running config or a show IP route to see the routing table. I had to shift my mindset from, rather than doing that, either using the console or using something like AWS CLI to actually see what the configurations are in the region that I'm in.

Alex Perkins: 51:40

Do you think your, you know your automation kind of background, if you will kind of helped with that transition as well?

Du'An Lightfoot: 51:48

I think automation did help me since that when you learn to automate, do you understand your networking? But you also think about the extraction of the network. Let's say, like your routing table, rather than logging into a device and doing a show IP route, you're actually sending that via an API call or you're actually sending that through some type of SSH client like NetNico or something, and so that kind of is an extraction layer from actually being on a physical device. So doing that actually compared me for the way I interact with automation in the cloud All right.

Shala @GiftedLane: 52:22

So this goes back to the other question where I said it would probably answer this. There was no mindset change for me. What do you mean? If anyone remembers, my endpoint devices as a network engineer working on prem was on ships. So I've kind of always worked in a space where I couldn't just go and like touch the equipment. So if I accidentally shut down a server, ad, something like that, I'm calling a captain or flying out to the ship to go turn that thing back on. So it wasn't a shift for me. I guess the only shift for me was learning where to do certain networking things on different cloud platforms. So you know where do you to go inside the page? I definitely know. I can't even say that because I was using CLI with the stuff on the ship. So yeah, it wasn't a big mind shift for me, to be honest with you.

Chris Miles: 53:19

And you've kind of leaned in pretty hard to the DevOps side of the house as well. Yes, how do you think that ties into it? Do you see that as a mindset shift? Obviously that operates a little bit differently than traditional network engineering.

Shala @GiftedLane: 53:31

Yes, first off, let's start with the fact that, if this is going to be a video I put out for my blogging stuff, I never wanted to be a DevOps engineer.

Shala @GiftedLane: 53:42

Let's start there. I had no desire to do it, I didn't study for it, it's just something I got put into at work and so every day was learning on the job. I have touched databases more than I wanted to and had to dust off my SQL skills. I've had to learn how to automate a bunch of stuff and actually truly understand what DevOps is, and I just came to a revelation like maybe December-ish this past December, where it finally clicked for me, as a DevOps engineer, my job isn't to do all the things. My job is to unblock my devs, teach them so they can go do the things, and then they teach me about their stuff, which makes it easier for me to support them. And that's where that infinite loop actually comes in. And I happen to be on a project where DevOps, like the culture, is amazing and we're actually doing DevOps, and once I figured out that part of it's not about me as a DevOps engineer doing all the things, but unblocking my devs so they can go do it. It made my life so much easier.

Andrew Brown: 54:58

Yeah, so I mean, my background is more of a developer, so my on-premise is more like what was it like before you used a cloud service provider? And back in the day we did not have all these more complex components like cloud networking. And I think that the largest hurdle for me to get into cloud or cloud service provider was actually to learn how to do basic IT networking, and once I overcame that, then I was able to jump right into cloud.

Alex Perkins: 55:27

Yeah, that's a good point Because I mean, the table stakes are a lot higher right, like there's a lot more of a baseline foundation of things that you need to learn like right off the bat, and I think that kind of ties into your hot take a little bit too. So for sure I agree with that one.

Andrew Brown: 55:41

Even more so now, like even when I came in it was it was just like I didn't want to even know what a subnet or cider blocks were.

Chris Williams: 55:47

I'm going to go back to the considerations about around monitoring. When I was living on-prem, when things were on-prem, it was easier to visualize where the failures could happen and to go trace them out. When you move up to the cloud, there's a layer of abstraction that can get in the way of good troubleshooting methodology until you actually understand how the things are fundamentally connected to each other. Once you cross that chasm, then you are able to troubleshoot and understand the differences in the monitoring perspectives of cloud versus on-prem.

Tim McConnaughy: 56:26

That's really good. I mean, that's a good answer. I would 100% agree with that. So much easier to find your blast radius on-prem than it is in the cloud and work back from that.

Chris Williams: 56:36

If you're staring at a rack and you've got your switches at the top and the router right beneath that and then everything else underneath that. Then you know where you're going to go.

Tim McConnaughy: 56:45

Yeah, that's really good.

Chris Miles: 56:47

Yeah, I mean, do you think there's a lot of pushback from people converting infrastructure from on-prem to cloud just based on that particular piece alone? Because, at the end of the day, slas are all fine and dandy but they can't necessarily save your business, depending on how catastrophic the event is. But yeah, do you see that being one of the major friction points for people as well, just that particular monitoring aspect alone?

Chris Williams: 57:18

So I've been party to a lot of cloud migrations. That was my first specialty when I went into consulting is getting people to consistently and with a lot of forethought and move systems up into the cloud. Honestly, that was the last thing that they worried about. The first things that they worried about were the sillier things, not sillier I mean. There were the things that fear and lack of knowledge of how the cloud operated was the first stumbling block. Getting convincing people to worry about monitoring and troubleshooting and understanding all of that was far further down the line just getting the person to even buy into the concept of cloud.

Chris Williams: 57:59

I will say that a lot of folks that are around my age were very reluctant to I'm not going to say believe in the cloud, but to be bought into it. I think a lot of it was just a reluctance to just pick up a new stack to try to have to. What I got to learn? Another thing I just spent the last 10 years of my life doing VMware stuff and data center stuff. Why do I have to do everything through an API now? So that was a much bigger hurdle than the actual real hurdle of trying to figure out why things break in the cloud versus on-prem.

Chris Miles: 58:30

Yeah, 100% Fear, uncertainty and doubt, and is always the scariest, especially at the tail end of 2020, which the tech treadmill was just like.

Chris Williams: 58:40

Oh, add another thing, add cloud, right they have more things network automation, cloud, python, all this other stuff, right Anyway especially when the interested parties are trying to inject even more fud into it Hyperscalers vying for land grabs among companies, on-prem service providers trying to maintain their relevance and continue selling hardware to each other. I always try to figure out where the money was going to land before I ask somebody a good question about something. If I ask AWS a question about whether or not I should move my things to the cloud, if they start getting dollar signs in their eyes, then I know okay, well, you might have a biased opinion about this. It's hard to get an unvarnished, unfiltered truth out of somebody that's getting paid by that company.

Chris Miles: 59:26

All right With that. We are going to go ahead and wrap up the episode today. I hope you enjoyed this. Hopefully this was a fun exercise to do something like this where we don't know exactly how it's going to come together, but hopefully it was a slam dunk. Once again, just want to say a quick thank you to everyone that's done anything at all to support the show in the last year. Whether that should be listening one time, maybe you never came back again, that's fine. Maybe this is the first and the last time you listen to the podcast ever. That's fine too.

Chris Miles: 59:59

Really, from the bottom of our hearts, really appreciate it, really want to grow the show and really resonate with the audience and talk to more people. That's the thing we really want to talk to as many of you guys as we possibly can and showcase things that you may not be seeing in the field or talking about, etc. We would love to get in touch with every one of you guys. We're looking forward to doing this again next year. Hopefully we have a bigger and better idea for next year. It's going to keep happening Every year. It's going to be a crazier episode. We're going to figure this out With that. Once again, thanks everyone. If you enjoyed this, make sure to watch on YouTube. One of the responses in the survey was that people were joking, that they didn't even know that we had a YouTube channel. I don't know if that was a joke or if that was a real comment, which it might be real.

Tim McConnaughy: 1:00:56

That was a sizable amount too. It wasn't one person there was a decent amount of people that don't know. We have a YouTube channel.

Steve McNutt: 1:01:01

We have a.

Tim McConnaughy: 1:01:01

YouTube channel If you haven't been watching us on YouTube, you can actually see our faces. That's probably why they haven't been watching Exactly.

Chris Miles: 1:01:12

Just blur out the portion of the screen where me and Tim are in. You can just watch Alex and you'll probably enjoy yourself.

Alex Perkins: 1:01:18

Or you can watch Tim's beard grow in and out every single episode.

Chris Miles: 1:01:23

That's a good point On this episode. Since we recorded all these at several times, you probably see Tim's beard constantly changing lengths between the videos.

Tim McConnaughy: 1:01:31

I don't have a clock on the wall behind me, so this is like the next thing.

Chris Miles: 1:01:35

Tim's going to start recording with a copy of today's newspaper next to his face every time. All right, with that we'll actually wrap up. Thanks again, guys. We'll talk to you next time. Bye, hi everyone. It's Chris. This has been the Cables to Clouds podcast. Thanks for tuning in today. If you enjoyed our show, please subscribe to us in your favorite podcatcher, as well as subscribe and turn on notifications for our YouTube channel to be notified of all our new episodes. Follow us on socials at Cables to Clouds. You can also visit our website for all of the show notes at CablesToCloudscom. Thanks again for listening and see you next time.