The Shift from Kubernetes and DevOps to Platform Engineering

00:00:00: Welcome to Signals, the podcast where we step back from the hype and look at what's shaping technology.

00:00:05: In each episode We talk with industry experts to surface trends question assumptions an unpack why things work The way they do.

00:00:11: today were focused not just on whats new but On the decisions in developments that got us here And What To Expect Down The Road.

00:00:24: I'm your host Michael Dowden Here With Me.

00:00:26: Today is Abdel.

00:00:28: Thank you so much for being here.

00:00:30: i think We've known each other for a while now.

00:00:33: Uh, we met first in Morocco and twenty nineteen I believe.

00:00:38: Yes yes been awhile.

00:00:41: So uh tell me what you've been working on.

00:00:45: Yeah Um so hi everyone it's Abdel.

00:00:48: i'm a developer advocate at google.

00:00:50: I have been...I think..so I've been workin on Kubernetes for quite a while, I think i can consider myself as subject matter expert now.

00:01:01: Mostly...I mean started working on Kubernetes about eight or nine years ago at the consultant because we have quite a lot of users on our platform using Kubernetes as tool and then over the last three years I transitioned to an advocacy role where I am talking with people how they actually should use Kubernetes more.

00:01:18: so yeah

00:01:21: So when you search for Kubernetes information, your name does tend to pop up from time-to-time.

00:01:30: I guess i know why!

00:01:32: So tell me a little bit about like what is the state of kubernetes today?

00:01:39: Is this something that you think all developers should be in and using if they're on the cloud or it's only for specific people working with?

00:01:49: where are we today?

00:01:51: Yeah, it's a very interesting... It's a good question actually.

00:01:56: I will take you down the history line and during this history where we're also going to talk about DevOps because they kind of go hand in hand together.

00:02:07: So Kubernetes is eleven years old right?

00:02:09: So two thousand fourteen was the first time in Google open source Kubernetes And the project now has eleven twelve years its premature Adoption through the roof.

00:02:18: A lot companies have.

00:02:19: most major cloud providers have some sort of offering around Kubernetes.

00:02:24: Today I was just looking into some internal stuff, about a hundred products on Google Cloud uses Kubernetes actually under the hood that people don't even see right?

00:02:35: And this question is like Is This A Developer Tool?

00:02:38: Have Been Kind Of A Point Of Contention Over The Last Couple Of Years and People Have Been Going Back & Forth Because There's Like a love-hate relationship.

00:02:47: And there's very simple reason, Kubernetes is an API driven system so as developer you are kind of pulled toward anything that has an EPI but it still infrastructure which most developers don't like right?

00:03:00: So its way to drive infrastructure using an EPi compared having to do infrastructure itself with Biscuits.

00:03:09: I think Kubernetes probably a way better of doing it, but is still infrastructure and you have to understand the IP addresses firewall rules, network policies all this stuff that most developers don't.

00:03:19: And frankly shouldn't have to care about them.

00:03:21: I think at the same time there was evolution of DevOps right?

00:03:24: Like DevOps has been around for awhile people talking about it they sort went hand in hand.

00:03:31: if you are developer just building backhand or front-end code For the most, you don't have to care about Kubernetes.

00:03:37: It's just there.

00:03:38: it just works right?

00:03:39: You have a CICD pipeline push code got get built and get deployed.

00:03:42: You can check out your code in production And then move on with your life Right?

00:03:47: but Then There is this new kind of evolution In the DevOps space called platform engineering which Is sort Of like The next generation way of looking at DevOps instead of considering more Like is This A role or an engineering practice Considering Everything that developers Have To Run On As A Platform?

00:04:05: and whoever takes care of the platform is called a platform engineer, right?

00:04:10: And so in that space I think there's still this clear line where developers for the most shouldn't really care about Kubernetes.

00:04:17: They should just assume...I'm putting an Assume on Code here In the sense it doesn't have to be there but if its' there It should just be transparent.

00:04:26: The most important thing i think That Platform Engineering as a practice brought into the table Is this entire concept an IDP internal development platform, where if you are doing Platform Engineering right what you're supposed to be doing is give your developers a platform that they can just shop in through a portal.

00:04:44: So they go and say like gimme database, gimme this, gme datgimme this.

00:04:48: Give me place to run my app.

00:04:50: I don't really care whether it's Kubernetes or not.

00:04:53: The state of Platform Engineering report which came up in twenty-twenty five about sixty percent of companies adopt Platform Engineering as concept.

00:05:01: actually have Kubernetes under the hood, right?

00:05:03: That makes sense.

00:05:04: And what an IDP does really is that it gives developers much higher level... If you think of Kubernetes as an abstraction layer over infrastructure and IDPs in abstract layer one-level above where they give you the Kubernetes infrastructure then everything else your app meets because your app doesn't live in a vacuum.

00:05:22: It needs database, certificates, authentication stuff like this.

00:05:28: So I think there's long way to say.

00:05:29: for most you shouldn't Unless you have a very particular case or very particularly use-case, which is... You are developing products that are tangibly and tightly related to Kubernetes itself.

00:05:41: And unless you're in these kinds of situations for the most it shouldn't matter if we were as developers.

00:05:45: So has lot's do with how this type product your building?

00:05:49: How you've architected your product and how you intend to roll out to customers?

00:05:53: I would imagine.

00:05:54: So if it's something where you're wanting to spin up infrastructure dynamically in real time as you onboard customers, then that might be something that is integrated directly API level into the way we build an application for example.

00:06:07: Exactly!

00:06:08: I can give a very good example here.

00:06:10: We have a lot of customers who are building tooling for developers so they're building things that developer use.

00:06:19: The example comes from mind.

00:06:20: there were companies today backend system for running Flutter code.

00:06:24: So Firebase like a platform, so not Firebase but something that's like Firebase?

00:06:30: What you are building effectively is something that is targeting developers.

00:06:33: You will have to deal with infrastructure anyway and for them Kubernetes gives them the API so they can provision the infrastructure without having to care about the underlying machine But then their developer or users shouldn't really care.

00:06:45: I just run command line.

00:06:47: it gets pushed through cloud.

00:06:49: Yeah,

00:06:52: that makes a lot of sense.

00:06:54: So how do tangential technologies like GKE and the service mesh stuff like that?

00:07:01: How does those interrelate with how you work with Kubernetes?

00:07:07: Yeah I mean GKE is just the product.

00:07:09: it's Google Kubernetes Engine.

00:07:12: we are not only one who has this.

00:07:13: EKS has Elastic Kubernetes Engine.

00:07:15: Azure has Azure Kubernetes Engine.

00:07:17: It basically manages Kubernetes right.

00:07:18: so Like same way to manage database or manage PubSub system, it's a managed Kubernetes.

00:07:25: So we manage the entire thing for you get an EPI which is standard Kubernetes API and then talk to it using KubeCTL command line for Kubernetes.

00:07:33: so that's like most basic things.

00:07:36: Anthos technically doesn't exist anymore in this sense as we started with it.

00:07:41: but what Anthos really is or what we call now GDC Google Disconnected Cloud or Google Distributed Cloud I don't remember.

00:07:51: It's basically a packaged Kubernetes that you can run on premise, right?

00:07:55: You think about it.

00:07:56: I have been in the industry long enough to remember there was time when we talked about Linux as an operating system and then we talk about Ubuntu as this solution.

00:08:06: So Ubuntu is just like an open-ended version of Linux so you can't think of GDC as an opening version of Kubernetes.

00:08:12: So Kubernetes is an open source thing And then you get package versions which are running on premise.

00:08:17: That something you know, if we take all the developers in the world and say that maybe ten percent of them will have to deal with communities then one percent or ten percent.

00:08:26: We'll never talk about what GDC is right?

00:08:29: And service mesh on another hand... That's actually a very interesting piece of technology.

00:08:37: I think has a future.

00:08:39: It's been slow to adopt because it solves problems at infrastructure levels.

00:08:45: most people again shouldn't care about What really it does, it solves the problem of how can you add intelligence to the network layer.

00:08:54: I can give a very simple example.

00:08:56: so let's say for example you are rolling out new version of your app.

00:09:00: You have the standard regular rollout strategies, blue-green, canary whatever you want to do and most times these are pretty complex like set up some sort of proxy to be able route parts with traffic to the new version.

00:09:18: And so what really service mesh as a concept have been introduced to do is solve these kind of problems, but using configuration.

00:09:26: So instead of having deal with the underlying infrastructure we're just configuring in YAML format to be able do these things on our behalf.

00:09:37: Canary deployment, blue-green blah blah and so service mesh.

00:09:40: I think the first tool that was introduced with Istio about like seven or eight years ago.

00:09:44: um That also came from something that Google was doing.

00:09:47: we've been doing for a very long time.

00:09:48: So we just took the same concept And we built something that looks like Istio.

00:09:52: in A very funny coincidence it followed The same history of Kubernetes.

00:09:58: so Kubernetes was initially released To manage Docker containers.

00:10:03: So Istio was initially released to manage Envoy, which is a proxy.

00:10:08: Which Envoy was built with the Lyft independently from the service mesh itself somebody I think it's been built in it and they were like... They had no idea that this thing called ServiceMesh existed until we open sourced right?

00:10:19: And so It is really layer on top of Kubernetes.

00:10:22: you can install some vanilla Kubernetes managed Kubernetes or on GDC or Anthos or whatever.

00:10:28: again for the most As a developer you're probably not gonna have to deal with it if It's part of your platform that it should be available for you in the IDP where we just click button and then you get canary.

00:10:41: But it's something.

00:10:42: That's probably your security team.

00:10:44: Your network team will have two, you know install managed and deal with.

00:10:50: so If you are a DevOps engineer or platform engineer how Would you say your life has changed over the last few years as a direct result of having Kubernetes compared to, I don't know older stacks?

00:11:08: Oh yeah.

00:11:09: Clearly going from writing scripts and writing declarative or writing configuration that's probably like the biggest change right?

00:11:17: you know, the old version of DevOps engineers which I think we can call system administration.

00:11:23: Right?

00:11:25: You've had to deal with Bash maybe Python or your way around it.

00:11:29: then there was a little bit at some point where new tools started emerging.

00:11:33: things like Chef and Puppet kind achieve same thing but in more server clients configuration architecture Kubernetes came to the picture alongside Terraform, which is around at the same time.

00:11:46: Which essentially went from a prescriptive programming language through declarative programing languages like instead of you prescriptively telling the platform what to do or just declaring your desired state.

00:11:56: so that's been probably the biggest shift for DevOps engineers in general and then I think started working a few years ago.

00:12:11: Most of them have started working in the cloud, so for most they probably don't even know that before firewalls were distributed there was just some box at the data center right?

00:12:23: Before load balancers were software driven and like an actual physical box somewhere.

00:12:28: I

00:12:29: remember those yeah here you

00:12:31: go!

00:12:32: That's not to shift as many experienced.

00:12:35: but What's actually I think is going to happen, go in forward.

00:12:38: Is like the reverse shift because there isn't huge now and that's a huge, it was like a trend then going back on premise people know how to go back to deal with physical boxes and physical hardware or physical infrastructure?

00:12:51: And so.

00:12:52: but I would say yeah, The biggest shift is the shift In the way you do things.

00:12:56: So programming two configuration driven Things are also maybe the shifts in the mentality of splitting responsibilities kind of layering your stack and being responsible for a layer that somebody else depends on.

00:13:11: And yeah, I mean...I guess now we cannot avoid it.

00:13:15: like AI is just impacting everyone in many ways.

00:13:20: then people want to admit or are comfortable with but Yeah!

00:13:25: It's changing all the time.

00:13:26: But they're just how industry is right?

00:13:29: Absolutely So.

00:13:33: Would you say that the shift off of the cloud is the biggest thing that's happening right now in this space or?

00:13:40: Is there something else that you see as kind of the next The next big shift.

00:13:46: So, you mean off-the-cloud back to our premise.

00:13:48: Yeah

00:13:49: I mean it is a trend That we are seeing.

00:13:55: I guess probably the loudest voice on this space would be DHH, the signal-thirty seven going from AWS on premise.

00:14:07: There are so many things to say about that particular... It is a great lesson right?

00:14:11: So whether you agree with what DHH writes and you agree everything like all of conclusions or let's a pragmatic look at things and you can like, look at it from the perspective of when you started going to cloud somebody should have pointed that this is the wrong approach in the first place.

00:14:33: That's just discussion for another time but I think that the point is yes there is a trend of shifting back on premise But there is actually very interesting thing that i am learning over.

00:14:48: Like when we say a shift back to on premise We're really giving it the wrong term in this sense.

00:14:53: that To a lot of people like me.

00:14:55: It might sound like oh, I'm moving off-the-cloud back to my own servers But in reality what's happening?

00:15:00: I think is I am moving off the cloud too a man and a server farm.

00:15:06: So i'm still getting managed infrastructure somewhere.

00:15:08: I'm just not getting the cloud version.

00:15:11: I'm getting A server that I pay for but I am NOT renting the rack.

00:15:15: I just pay somebody, and then they install the racks for me.

00:15:19: And they give me servers that don't have to even physically touch

00:15:22: them."

00:15:23: Right?

00:15:25: So it's sort of like...I guess if you go back to old-school classification of pass EAS SaaS, we're going back to a platform as service kind situation.

00:15:38: Yeah this is clearly something noticeable.

00:15:44: one of the biggest signals is like all the cloud providers today are interested in giving you a piece of software to run on premise because they're seeing that trend and there's business to be made, right?

00:15:54: So that's one them.

00:15:55: And then I think... Like i'm gonna shift little bit toward future parts of your question.

00:16:03: In the future what we'll see more?

00:16:06: if we continue going towards those AI models everybody wants to run now being either A too expensive or be too hard to obtain the hardware you need to run on, then we are probably gonna go through a situation where people will just want to buy the hardware and do it themselves.

00:16:23: Right?

00:16:24: Or for privacy reasons, or for sovereignty reasons, all of these reasons that people wanna run on prem.

00:16:29: So I think this shift is continuing to exist...or will continue to exist whether it will accelerate or not so that remains to be seen.

00:16:37: That

00:16:39: makes sense!

00:16:42: For those who remain cloud native, is there a next step beyond some of the Kubernetes?

00:16:56: Obviously I've been working in the serverless space for quite awhile and Serverless benefits a lot from Kubernetes.

00:17:00: I don't have to touch Kubernetes off to worry about it but that's how realistically i believe a lot of providers are making serverless options available to us right And so I do not want to be on the server less space.

00:17:11: I dont' need to worry because DevOps is very light in that world.

00:17:17: Is that the future or is there something even a step beyond that where we're going on cloud platforms?

00:17:25: Yeah, I think it's... There is space for serverless to remain and maybe grow over time just because they will be more people who build apps but don't want to do DevOps right?

00:17:40: Serverless.

00:17:43: I am the biggest fan of Kubernetes.

00:17:44: I would be the one to say, start with serverless like you don't have to start Kubernetes.

00:17:48: and then if you get mature enough where do you need Kubernetes?

00:17:52: Then maybe he can stop thinking about it right.

00:17:55: what i think The pieces of technology that's I'm actually keeping an eye on is It comes in multiple names.

00:18:06: I call its edge workers So that essentially being able to run some sort of lightweight code closer user.

00:18:12: So that would be things like CloudTrain workers for example, which is essentially a serverless platform except it doesn't run in data centers.

00:18:20: It runs on the node closer to user so you can do pre-processing or maybe all processing unit.

00:18:26: if the process is very lightweight then its great.

00:18:29: I was watching an episode of I think Teo TTG like a very popular YouTuber who has this platform called T-Tree Chat, which is a wrapper around multiple AIs that allows you to talk to multiple Ais with single subscription.

00:18:44: And he was talking about how they moved off next GS too... I don't remember the name of the framework there running on right now but they have been moving toward these new frameworks and then in the process of moving They've been relying a lot on edge workers To deliver You know The JavaScript assets bundles to the user much faster Right?

00:19:04: across and at the edge.

00:19:06: And I think that's going to continue growing, That portion of the market will continue growing right?

00:19:11: So that for sure one serverless clearly is going to grow in.

00:19:17: The biggest indication why this is gonna continue growing Is there now interest from big cloud providers To providing you with accelerators on top of serverless platforms.

00:19:27: Cloud Run has L four GPUs.

00:19:30: there will be more type of GPUs coming.

00:19:32: That's because, you know as developers I want to run a lightweight LLM or an LLm in general on my serverless platform.

00:19:38: i don't want too...I don't wanna be strong-handed go down into the Kubernetes world!

00:19:43: I just wanted to continue working it in my space and.. ..i think that is also this.

00:19:50: other portion of industry which probably we dont talk about that much but i dont see that much is edge workers But not in context of Cloudflare.

00:20:00: You know, I am a big fast food chain that needs to run workout workloads on in the store.

00:20:06: Right?

00:20:07: Like I need to run checkouts and point of sales in the story itself And you don't want to mention any names but like insert fast-food chain.

00:20:17: Think about it.

00:20:18: as you walk into your favorite fast-foot chain then talk to a kiosk machine maker order.

00:20:24: That machine is not talking to a back-end running cloud.

00:20:27: It's probably talking to server run in the same store where you are, right?

00:20:30: And there is very popular American first food chain that started doing this and started blogging about it.

00:20:37: over time called attention of a lot providers now more offering space as well.

00:20:47: So with kind To some companies doing more on-prem whether that's because they want the privacy or the control Or because like the fast food situation you need Maybe a server a little bit closer to the actual point where the users in are facing with it.

00:21:10: This feels Like It's A Little Bit of a Pendulum Swift Shift Right, We Get These Pendulum Swings In The Industry Where Everything's as close to the user is possible and everything's out in the cloud And we get these swings back-and-forth.

00:21:24: this feels like another one of those pendulum swings.

00:21:26: To me, would you say that?

00:21:28: That's correct.

00:21:30: Is this What?

00:21:32: what does the next swing look like once we get through this like water?

00:21:36: What are some of the reasons?

00:21:37: you might see somebody making choices for the future?

00:21:44: Like what other than decision points that you see people making?

00:21:48: Yeah, I really like this analogy of pendulum because um i think that the future would be That that pendulum eventually will have to find equilibrium at some point and stabilize right?

00:21:58: Right.

00:21:59: Um And in The process of destabilization happening.

00:22:03: I think what's gonna Happen is we're going to get to the point where realize as an industry which workloads run best closest user on Which ones run best closer to them to the data center right right and which one runs in between Right?

00:22:17: And it's not like, It might sound to the audience that... The way we're describing is kind of extreme.

00:22:24: Its binary either you are closer user or they're closer to the data center.

00:22:27: but that's not a case today because your in between right.

00:22:31: I mean best example i can give you when you watch YouTube.

00:22:34: You don't stream any youtube video from a datacenter and you're streaming it on a node closeer.

00:22:38: But for this the technical term CDN content delivery network doesn't necessarily has to be physically within few miles from you.

00:22:48: It can be a hundred miles for me, it's just that the latency toward that node is much lower than the latency towards data center in different country right?

00:22:58: So I think we are in this process where we're trying to figure out Where Is The Best To Run The Workload If We Want To Go Even Extreme Whether Its On The Actual User Device because that's really as close to the user as you can get, right?

00:23:17: Absolutely.

00:23:18: In the browser or all the way back in a data center or anywhere in between and I think this pendulum will continue swinging back-and forth.

00:23:26: us new technologies are maturing has optimizations are being introduced to make things faster better cheaper like Etc etc.

00:23:37: And then at some point we're gonna get We're going to get through some sort of equilibrium where we exactly know, depending on the characteristics of workloads.

00:23:47: Do we run it actually in one side?

00:23:49: In other side do we run all the way between them?

00:23:51: and what I want really to like this is just because its fresh in my head.

00:23:56: i've listened to the pragmatic engineer.

00:24:00: Gregori probably knows him And he did an episode recently with the VP at AWS.

00:24:08: It was amazing right.

00:24:11: This particular episode the VP, she went really deep into details of how history works.

00:24:17: And that's like a completely different... Like we're talking about planet scale systems.

00:24:22: it is insane right?

00:24:23: Right!

00:24:24: I think there will be in my opinion from pure technical and engineering perspective more need for people to focus on these kinds of use cases because clearly Running things in data centers is not enough anymore.

00:24:40: We need more capacity, we need more storage and compute And there will be more focus on running this large scale.

00:24:49: You know ten years ago we probably used to say that large-scale systems are for the big scalers like Google, AWS, Microsoft.

00:24:56: I think today if you're doing a large language model or media whatever your'e doing foreseeable future.

00:25:06: You're gonna have to design a system that runs at very large scale from scratch and you need people do that, right?

00:25:12: Right

00:25:12: exactly

00:25:13: for sure!

00:25:15: So I'm really happy... well first of all this is really fascinating like the S-III global scale is unbelievable.

00:25:22: oh yeah it's interesting.

00:25:23: iIm really happy brought up the running on the client device.

00:25:27: That's something I've actually been doing for a really long time and am huge advocate for because sometimes that is just best place to do work.

00:25:36: so my question, when you look at kind of an industry wide approach Do You See A Future Where We Have A Framework?

00:25:46: That Kind Of Helps Us Build An App And Deploy It All The Way From data centers to the edge, to the client device and the framework kind of helps decide where any given job runs based upon lessons learned over time.

00:26:04: Or do you see that as more a problem?

00:26:07: That we're gonna use AI to solve at the time of development versus in real-time is part of platform or product?

00:26:19: Yeah so Whether there will be a framework, I don't know.

00:26:22: Because i'm more like a backend person.

00:26:24: so... ...I don't really know much about kind of front-end or end device type frameworks that are specifically designed to run things and workloads on end user devices.

00:26:34: What I know is that it isn't evolution in the primitive APIs under the hood that enables these types of workloads.

00:26:41: So example can give you the most recent one which we probably know Is the web AI API inside Chrome.

00:26:48: That's essentially running AI workloads on the browser, right?

00:26:52: So that's not a framework per se.

00:26:54: That's more primitive.

00:26:55: it is available inside the browser and then you have to build on top of it Right.

00:27:01: so I think like if i can put my speculation hatch here...I think this would be hard To Build A Single Framework That Can Figure Things Out.

00:27:11: I

00:27:12: Think What Is Most Likely To Happen Is That These AI Tools Will Become More And More Intelligent And with enough training and enough data, in enough use cases that they get exposed to.

00:27:24: They will be able help developers figure out what's the best way to do certain implementation right?

00:27:31: Where you can ask their large language model describe your spec driven development.

00:27:36: You're describing constraints or describing business requirements maybe are describing latency requirement technical requirements and then the model tells okay well I think for this portion of application This would be better for this other portion of the application.

00:27:49: These will be better,

00:27:50: etc.,

00:27:50: etc.. So I think that it's one of things we really like about our industry is almost everything is built on open standards.

00:28:01: so were not at the mercy a single provider to provide us with their framework!

00:28:06: Right?

00:28:06: So i think thats what.

00:28:07: when they realistically happen maybe frameworks are trying solve these problems and then Some will stick, some will disappear.

00:28:18: and as kind of we are progressing or learning.

00:28:21: We're gonna get to a point where either we have very clear understanding depending on the requirements for workloads... Where do you do it?

00:28:30: That's one!

00:28:31: Or we are having an AI that can tell us this right And I don't know if i'm going to anticipate your next question.

00:28:41: This actually makes me think The future is more about as an experienced software engineer Do you have the knowledge to be able to make the right decision?

00:28:51: About this kind of things, right sure and And I'm gonna stop here just one last comment Just if people are actually still wondering what the future looks like.

00:29:02: I was in a call today where we're talking about FTP servers.

00:29:08: I wasn't calling with customer.

00:29:09: they are using FTP server so You know, just chill everyone.

00:29:18: That is a fantastic context honestly so it really some.

00:29:24: I guess that changes my next question because i want to look backwards Just A little bit.

00:29:30: if you're not cloud native today If you're Not using kubernetes if your NOT Using all alums like what are the alternatives To Some of these things that are happening in the cloud native space today?

00:29:46: I mean, so there is a huge part of the industry which still running on virtual machines.

00:29:51: The virtual machine industry does not just disappear... ...I will be willing to bet your again.

00:30:00: insert virtualization company here!

00:30:03: I think everybody knows who i'm talking about right now.

00:30:07: I would be well into bed but they're making big bucks or big parts of their business.

00:30:11: They might be wanting to push Kubernetes down the throat of their customers, but like most customers just don't want it or don't understand it.

00:30:21: Or you know how it is?

00:30:24: Like I would say that biggest challenge is change.

00:30:28: as long as we can convince people to change That's when you can push technology forward and if he can't convince people who want to run on things they're familiar with.

00:30:38: virtual machines have been around for a very long time.

00:30:40: We know how they work, so if it ain't broken don't fix it right?

00:30:45: So I think the alternative is virtual machines.

00:30:47: clearly that's going to stay around for the foreseeable future.

00:30:52: i would say large language models.

00:30:56: really you don't need them.

00:30:57: If You Have People Who Have Experience like experience engineers Don't Really Need Language.

00:31:05: they can just look at log and tell you what's broken, right?

00:31:09: You and I could look at an error.

00:31:11: And though it's DNS rights is very simple.

00:31:13: so yeah So i think that there is still a big part of the industry which...I don't want to..i think The word would use Which i dont know if its the right words immune in way to this like advancement.

00:31:28: Because They Just Don't need them.

00:31:30: um they are running on systems That already work And already function properly and they don't even need to know.

00:31:36: They don't need cloud native devops, they do not need Kubernetes or LLMs just like continue doing the things that are going in.

00:31:42: what's the future for these workloads?

00:31:44: I think it depends.

00:31:46: um i believe That The biggest Driver toward innovation For any business It doesn't matter really What a business is is competition Like the moment when you have an established company Doing something and Don't want him To change not wanting touch Any of their systems, whatever.

00:32:03: not wanting to innovate.

00:32:05: It's wherever innovates means you have a competition.

00:32:07: they will be forced to innovate right?

00:32:08: So yeah I think people would continue using the stuff that we've been using already.

00:32:16: so

00:32:16: there is company i know of who still uses FTP for files and as an AS-Fourhundred their primary backbone Always is a good, I think refresher for us as engineers.

00:32:34: As architects platform engineers to take a step back and look at the whole space that industry lives in.

00:32:42: And as much as we might be super excited about being on the cutting edge Maybe That's not where The company that were working For Is At.

00:32:52: maybe thats Not even Where they need To Be at any given Moment In Time.

00:32:57: Yeah Yeah, clearly.

00:33:00: I think it's good too and again i'm going to go back to that podcast with the AWS person.

00:33:08: she used this term that I found very fascinating.

00:33:11: um It's To Be Technically Fearless But Respect.

00:33:15: What's There So Clearly She is Talking About In The Context Of S-Tree Because S-tree Is A System That Is Sporting Thousands of Businesses And it is clearly not a system that you just want to change, right?

00:33:29: Like whatever changes introduced.

00:33:30: It has to continue working because there's people relying on this.

00:33:34: and um To your point for wherever people are working maybe That's actually the core of The resistance is just people are afraid.

00:33:44: if we introduce changed things will break.

00:33:46: So If you approach with empathy and respect toward what been their have been walking to suddenly start working better, right?

00:33:58: And then maybe also like thinking outside the box being technically fearless.

00:34:05: Whatever that means in a sense is you can just come in and look at a low hanging fruit, a pain point something that you can help solve for the department industry organization whatever it is.

00:34:18: Then at the same time not really introduce as much change to wherever exists.

00:34:23: so yeah I think That's a skill that people have to develop.

00:34:27: I would say

00:34:29: For sure Okay, so Based upon this conversation.

00:34:37: What would you say is the number one takeaway?

00:34:40: You want for the audience today like what When we look at cloud native will look at Kubernetes.

00:34:46: What's the big take away from your perspective?

00:34:49: Yeah So I will give you two, because one would be more forward-looking and another backward looking.

00:34:57: The way of looking at things is to somehow anchor yourself in the basics on how technology works.

00:35:07: as an engineer?

00:35:08: That's probably the best way that you can immune your ways by building foundations for how stuff works right now.

00:35:18: You know how to code, you know how design.

00:35:20: How to architect all that stuff like.

00:35:21: it's very well and then so that would be the biggest takeaway right?

00:35:28: And for the way forward I think people have to keep in mind is except maybe every few ten or fifteen years of our industry we're not really introducing much new inventions.

00:35:43: We are just building abstractions on top things which existed.

00:35:47: Those large language model didn't just happen three years ago.

00:35:50: They have existed for a very long time, right?

00:35:53: I think that the biggest word i would use to describe our industry is the ease of access.

00:36:00: What technology allows us do Is build abstractions That make things easily accessible To people.

00:36:06: so it's a lot of people.

00:36:07: It might sound new but two People who've been around For awhile Right?

00:36:11: Yeah.

00:36:13: And so whether that's Kubernetes, whether it has containers with that service mesh where there are those AI is just like we're building on top of existing stuff and as long as you have a very good understanding for the foundation how things work anything else comes after should be easy to understand.

00:36:30: totally.

00:36:31: yeah this fantastic advice thank you

00:36:34: awesome Thank You.

00:36:35: So I noticed That you just launched your YouTube channel.

00:36:41: Yes.

00:36:44: So that was a great video, so very cool!

00:36:48: What is new in your life and where can people reach out to follow you?

00:36:56: Finally I have my own blog on the Internet now.

00:37:01: It's abdel.dev site and you can find all my social media, on YouTube.

00:37:08: The first video I published in YouTube is a video that came two years later after i recorded for the first time.

00:37:14: but finally it's some personal branding for developers.

00:37:17: It kind of niche in the sense if your developer trying to make name yourself online.

00:37:22: how do go about it?

00:37:23: And covers things like social medias, content creation.

00:37:28: You know this stuff.

00:37:30: Otherwise yeah link.

00:37:31: then twitter where i hang out lots.

00:37:35: What's coming?

00:37:36: I am actually interested in Exploring like part of this YouTube channel that they created.

00:37:41: Um, um, I'm planning on exploring it further by creating content Talking specifically about foundation technology like covering things into foundational level explaining basics off stuff.

00:37:53: So yeah hopefully there will be more contents coming out There and i'm already working on some cool stuff.

00:37:58: And then we'll also continue doing interviews because I really enjoyed doing those.

00:38:02: And I think the other party is the Kubernetes podcast.

00:38:05: We continue to work on with my colleague for Google.

00:38:08: So yeah, either my blog or the podcast

00:38:11: fantastic Abzel.

00:38:13: Thank you so much for being here.

00:38:14: I appreciate the conversation as always

00:38:17: Thanks.