How'd you like to listen to dot NetRocks with no ads? Easy? Become a patron for just five dollars a month. You get access to a private RSS feed where all the shows have no ads. Twenty dollars a month, we'll get you that and a special dot NetRocks patron mug. Sign up now at Patreon dot dot NetRocks dot com. Hey Carlin Richard here. As you may have heard, NDC is back offering their incredible in person conferences around the
world. NDC Porto is happening October sixteenth through the twentieth. Go to Eddcporto dot com to register and check out the full lineup of conferences at ndcconferences dot com. Hey, guess what it's dot net Rocks. I Carl Franklin, and I'm Richard Campbell and we are still at the what is it the Developer Festival in Copenhagen, in Copenhagen, Copenhagen. And I don't know about you, but you've been here longer than I have. You and Stacey were here
for a while. Yeah, we kind of We got it on the weekend, so we got a chance a little touristing. Yeah, I didn't do any tourist things. I went. We went up the helsing board to the castle that supposedly was where Hamlet is set. That is so cool. Yeah, it's not that big of a big castle will be big enough for a big whiner to be end up written about by Shakespeare, but you know, just a whiner. That's funny. And then we took the ferry across to
Sweden because that's something you can do. That's cool. I've only taken the train. Yeah yeah, but they made me put it back, so that's what happened. That happens anyway. Any Telbosto is here, but before we talked to her, we're going to spin the crazy music for better no framework, Go ahead, all right, man, what do you got? Well, this came across my desk a little while ago and I thought it was interesting and I tried it. So the latest Adobe Premiere Pro or Adobe Premiere
for Video Editing has this new workspace for text based editing. Interesting text based editing. So what you do is you have to tell it to transcribe the videos that come in. And it's great if you just have one person talking, right, It's not so good if you have multiple audio tracks. And video tracks that are synced up and stuff. But here's what it does.
It transcribes it, and then you go into this workspace and you see the sentences on the left, and you can highlight part of the sentence what they said, and it adds that to the timeline. Interesting, so you can do your cuts and edits and stuff just by highlighting the words. And now you've got jump cuts where you're just highlighting the stuff that they said that you
want, yeah, and none of the stuff that you don't want. Well half the time, that's where you're lining up the cut anyways, just as someone starts speaking the word, right, although I do like the other voice starts speaking, screen fades in afterwards. Yeah, yep, yeah. But but on theline you can you can mess around with it, for sure.
But this is a good way to get stuff on the timeline. My problem is, you know, when we do videos for my other podcast, it's great if it's just me, but if there's two people talking, now you've got two video files, two audio transcripts, and you can't sink them like there's no you. You basically do jump cuts, you know, and then
they get out of sync, and now you're screwed. So, like I said, it's really cool for and I usually don't push you know, commercial products and things products, but but you know, anybody who's doing serious creation, you know, creator content has the Adobe Suite probably and you know this is just one more tool you can use. Awesome, and it is neat that they are working on things. Yeah, for a long time it was
easy to say Adobe was where good software went to die. Yeah. I was particularly angry with them about their you know, the pdf thing, right, I mean the Adobe pdf reader is like a virus. Yeah. At one plague it was a plague and you couldn't get rid of it. And now you don't need it. Of course there's the browsers to show PDFs. But but yeah, no, I don't want to subscribe to your thing, and shut up. I don't need a new update. And they Adobe pdf it's a pdf reader, just stop it. Stop. But let's face it.
You know, Photoshop, Premiere Pro Audition. Yeah, addition, these are excellent of the best of the best, and hopefully they keep them healthy. Yeah, they grab Figma too, I think I think they did. Yeah. Yeah, for better or worse. That's it, all right,
man, that's cool. Yeah, who's talking to us today, Richard, I grabbed a calm on top of the show seventeen twenty three, the one we did back in January twenty twenty one with our friend Tom Kirkhoff when we were talking about containers and Azure. So naturally we talked about Kupernetes because it's kind of unavoidable, and Karthur kaine Vic had this comedy says they take this is a great show with great insight. I'd like to add my two cents
about when to choose Azure functions versus Kubernetes. I had the same question where I needed to run machine warning workloads and Azure, And of course this is a couple of years ago, because let's face it, there's a lot of ways to do machine learning workloads these days. That's changed too, as your functions have a time out of about ten minutes, and even when you choose a premium Plummer app service plant has the limitation of not being completely serverless and
a limit of only fifty to two hundred instances at a time. By choosing Kubernetes, I was able to spend more nodes and also tape the pods to different high CPU nodes. So I think he's actually assigning them to high end nodes based on the job without a time up problem. I need at least five hundred nodes spinning up to five thousand pods to complete the workloads faster.
Wow. Kubernetes is not only completely serverless. When you use the virtual cubulet, you do need to pay for at least one minimum node and a load balancer. If you're running on one, you probably shouldn't be us Kubernetes anyway.
The whole point is scale, right. The main point is using Cubernetes, using Helm you can divine the pods, memory and CPU, which is not possible in the Azure functions for variable work Yeah, and Azure functions tend to go to sleep and some lucky customers kind of hit the jackpot and have to wait for them to wake up. Yeah, but you know that it makes the point. Functions is so abstract. You don't get to specify the
performance of a funtion. That's not a thing. Where Kubernatse is that tear down where it's like, oh no, I need you to make this thing with lots of muscle, because when I'm about to check at it's going to take a while more control, palay, So yeah, you trade some complexity or some control. So Cartha Kenyan, thank you so much for your comment and a copy of the US Dacoba. It's on its way to you. And if you'd like a copy of the US Ducobea, I write a comment
on the website at dot net rocks dot com or on the facebooks. We publish every show there, and if you comment there and a read on the show, we'll send you copy of music goby and you can follow us on Twitter or ex or whatever the hell they call it these days. That's fine.
But we're both on Mastodon and I'm on Blue Sky now too, so because I have so many places, you can just go to Carl Franklin dot com and right there at the top you can see on my social media all the media is yeah, and you are aware Combrichecampbell at Macedon dot social. I'm actually pretty much reach Campbell every everywhere. Not hard to find me. Yeah, oh good, follow us there, Yeah, send us a toot, send us a tweet, whatever makes you happy. Okay, let me
introduce Annie. Annie Talvasto is an international technology speaker, CNCF Ambassador acronym. What does CNCF stand for Cloud Native Computing Foundation? All right, you're an ambassador there, an Azure MVP and specialist in Kubernetes and open source. Annie hosts and produces the Cloud Gossip podcast and has been a co organizer of Kubernetes
and CNCF Finland meetup since twenty seventeen. She has worked at various tech companies, from cloud startups to enterprises, and spoken at tech conferences on multiple continents, including Kubacan. Do we say cube con Cubcan? I can't remember whatever you want to say. I usually say cube Con, but I everyone uses different It makes me happy. Con Cloud Native Con, Microsoft, Build and Ignite, NBC, Casey, DC Global, Azure, Future Tech and more.
During your career, she's spoken to more than thirty thousand developers at user groups, meetings and conferences. Wow, that's a lot of pizza. That's a lot. Actually, there's a lot of Harry it is. It is a lot more than that. Nowadays, I'm still calculating. At that point, I was like, whatever, it's something. It becomes a silly number. Yeah, exactly. So I just got to say this. My my mother's family comes from the Swedish part of Finland, about the Finish and the
Swedish in me. Yeah, that's cool, that's good. So I'm gonna get borshborshed not jiggy, really, I don't know. Okay, tired, I'm jet lagged. I have had more coffee than water today and I'm still tired. There you go. So what are we talking about. We're talking about kuberneties and we're talking about Tuli, yeah, or just the whole We're just going to have a conversation about kubernetes. First of all, what do
you think of the comment? The comment? I think actually that's at the gist of like the stuff that I've been thinking about recently, Like there's multiple levels that you can kind of think about communities nowadays. You can obviously think about it in a very practical terms of like, Okay, what tooling do you choose right now? Or how should you define or like how should your
teams work? And also obviously on the level of like what are the trends currently within what tools are teams choosing, or what are the different trends in
like how do organizations organize themselves and whatnot. But then there's the other level of think about communities or like infrastructure management in general, which is like what's the next thing after cubunites, or you can also usually think about okay, what was before cubunities, which can usually yeah exactly, and you can then kind of usually because I think there's really good sayings around like to know the future, you have to know the past or whatnot, so you can think
about those kind of things. And I think the comment is actually at the really good center of you know, cubunits is essentially right now that have fucked up way to do infrastructure and scalable exactly. Worried reading this comment not to chuck Karthy Kayan under the bus, but like he run it two years ago and this is still pretty young tag. Yeah. But also now I think all the people who were very early on in communities, they're already not moving
on. I don't want to say that there was moving on from cubunities, it would very much stay. But I think there's going to be another abstraction level added on top of communities. And as you can see, there's functions, there's server lists and whatnot. What's likely going to be going kind of exactly not next, but usually you can see in the past the technology goes
from abstraction level to the next construction level and whatnot. And I mean, I know it's about talking to folks inside of Microsoft's like as your functions is runs in containers. Yeah, you don't own those containers. You have no ability controls it is. But that's how they do it. There's no magic there. They're just abstracting it away from me. I think there's like an AI version that can control Kubernetes in the future coming, for sure, something
you can just talk to. Yeah, for sure, I would Star Trek Enterprise computer exactly. I think for sure. Obviously it remains to be then seen what goes mainstream, what's get adopted widely, and what's actually at the good balance of like okay, as you said, there was the challenges there around you know, the standardization and then the kind of flexibility to do what you actually need to do for you your case. So there's going to be
obviously questions around that. But I think I think in general, the next generation, particularly with this current like AI and Amlovs hype, I think someone's definitely going to go for it. Yeah, five millions exactly. And the
point being you don't need to know that number. Right when I think about a machine learning model sitting over top of a of a container infrastructure, like again going back to the comments, Hey, you want this level performance, so I'm going to tweak the shape of these these containers to prease their performance
and make more of them as necessary. That's getting beyond elasticity and it's some really intelligent behavior on how to optimize an infrastructure to get to sort of a a contractual deliverables, like we're gonna keep all these transactions under a second, if that means I have more instance, I have more instance. That means
I use higher levels of compute. I use higher levels of compute if I can, If I can have a natural language processor that will write my yamal for me, that's going to be a happy day, because that's going to be a happy thing except for that part where to miss this one indent. I like that. That's really great. And also look at the stratum now in the past couple of years of the Kubernety services. Yeah, so you're not standing up your own vms to host communities anymore. You kind of got
this pre configured piece. Yeah, very much. So, Yeah, that has been also, I think part of the abstraction journey for communities in general, because okay, if you want to do playing communities, you know, go for there's a lot of material, there's books. Community is the hard way, but I think there is it's in there already. Kubernity is the hard way, and you're not you're not saying that casually. Yeah, there is a hard way to do curriers exactly, So lock yourself out, yeah
yeah, yeah, and that is the actual book name. Also then obviously, like if you want to make your life easier, and also this is the standard nowadays also is to go with the managed service or managed community's route, which is usually can take the form or AK S, G E E S one of these from the hyperscalers or there's obviously a lot of local ones as well as smaller ones, or a combination off you know, managed service
with consultancy or service providers then helping you with some of the stuff and whatnot. Then obviously adding all that ons that you need on top of the managed services and whatnot. But I mean it might might be a bit of a contradiction of terms, but is there an on prem curbunities managed service like they a set of software tooling just to help you the same way that these managed
services do in the cloud. Well, the usually the premise of all of these managed services is with like the premise of communities, is that you should be able to use it on prem as well if needs right. I mean that was the whole claim it won so to an extent you should be able to do it normally. But there are also then kind of add ons and extra services that you can use there as well. I would say the likely
open shift, for example, has quite a good serving there. Like I work for a company nowadays that we actually somewhat specialize in providing open shifts on premises, right, which is essentially a version of communities on premises, for example, for highly regulated industries that need the on prem se. That's the red hat product, yes exactly, okay, but it's cubernatores under the hood. Yeah, okay. Now that's interesting, and it's just sort of recognizing
that there are ways to approach this. There's different layers of tooling on that. Of course, we immediately associate red hat with Linux stuff, yeah, but then in general you're making Linux workloads in kubernety containers. Yeah, I know we're supposed to be able to do it with Windows. I just don't see anybody doing it. Yeah, there's always often it's always to be honest. Uh, it's the word, the sentence that we all know. It
depends. Yeah, well, so like exactly what you need or what's to set up, and what's the kind of mix of services or products that you need, And it depends on what's the best and what not. But there's options out there. Yeah, there's no one right way for any of this. Yeah, I I feel like that the advantage the reason Kubernety sort of won in the orchestration wars I Air quoted that because we had DCOs and mesosphere
and like some of these other things was the ecosystem. Yeah, And I wonder if that's just like it got to a critical mass, it was now worthwhile to build better tools around it. It just became easier. Though, I would argue when Azure and Amazon made Kubernator services and it's like, and of course GCP was there first, It's like, okay, well game over. Ye You'll think maybe because the people at Google are freaking smart, smarter
than the other people and made a better product. I mean nominally that guy is Brendan Burns, who these days works for Microsoft right, and the last conversations we had with him, he's kind of thinking beyond containers too, Like that's a smart guy. He's always thinking about the next letter of problems.
That's where they're early on the folks at the moment. Yeah, but there is honestly a lot to do with the current set of tooling on what planning communities as well, obviously because as kind of mentioned here already, it is the way the containers are managed at scale, So there's a lot of things to consider there as well. For sure, what's your usual set of tools if you're setting up at kuberness It's like, I mean I can think of
Helm, yeah, but I know there's dozens. Well that's always depends again as well. Yeah, it's a bit of a bomber to keep repeating it, but truly I think, well, Helm is I think a really good place to start always. I think nowadays if you are moving into community circles, Helm and communities are quite often like inseparable, like you kind of don't do any any more than you then you develop in dot net without using new get right, like it's the package manager. You use a package manager,
your life will be better exactly. But there was like a few like I've just two part in like a quick interview, like you know, on these kind of rapid fire questions. And then someone did like within that interview sets it that you know, they wanted to sunset Helm and I was like, oh, I need to like find out more, like what's what's the reasoning here? But but there's always some hot takes and whatnot. So who knows
why? Yeah, because there's another one of those things that's got the network effect, like the fact that the package is all route through Helm, Like why would you try to disrupt that? That's a hard thing to get into place in the first place. Yeah, And I guess there's usually like there's new services and there's like if you look at the CNCF offering but not the
SENF. The foundation has like a lot of projects, open source project underneath this umbrella, right there is pretty much like if there's one way to do something, there's definitely another project to do a bit in a bit of a different way. Like there's gazillion ways, and there's stuff that that divides opinion that that for some cases it's really good in some cases, then some people find it lacking and whatnot, like cross Plane and whatnot. So there's a
lot of things you can choose from. So I would say that likely, yes, yes, I would say if you're starting your communities journey or midway point, Helm is probably a really good companion and good package management. But likely there's also a said before it depends, there's probably something else somewhere that may be really good for your specific scenarios and whatnot. Where do your tools like Terraform and Polomi come into play with Kubernetes, Yeah, well they are
infrastructure's code which has done a bit of a different thing. Well not, you can kind of use it in companion with kubunities. But if you are looking for as a code like project writer product that is more communities focused, then you would go with cross Plane likely, which is a project Yes, yes, yes, it's by a bound if I remember correctly. Yeah, cool, and so it does does what those things do, but it's more
Kubernates focus exactly. Yeah, well that's cool, rather than go out to another I mean there isn't a Kuminators plug in for Terraform, So I mean if you're used to working terraform and you're moving into cumuinators the plug it means you don't have to change. But again there's all this whole. If you're starting from scratch, why would you do that when there's more cumunaty centric tool
in front. That's all a command line anyway, right, you know one you know, and cross plane is relatively new, and you know there's nothing bad with new tooling and there's nothing inherently like unsecure or whatnot. But that is always the terraform is more of a standard, so that's always. Then you have to consider these things if you are making infrastructure traces. But I think, I think if I remember correctly, don't quote me on it too
much, but a thing cross plane is incubating level at CNCF. So CNCF has different levels for project materity. They're incubating, which means that they are not graduated, so not fully fully you know, full fledged for every scenario in production, but likely at the incubating level you're quite safe and good to go on in the CNCF. In June of twenty twenty, it is in the incubating levels. Yes, you called it correct, And what does that
mean to be an incubating level. So projects if well, they can start at any level. Obviously they come outside the foundation, but the life cycle. If someone were to start a new project, they would probably you know, start at first and have some level of idea and traction or like you know, something out there, then they would probably join in the sandbox level, which is like that that's your intro exactly. We just figured out what we want to be when we grow up. So there's a lot of sandbox
project nowadays. I think last time I checked, probably talking like seventy or so, and that's a tier that got introduced few years ago, so it didn't exist originally. But that's essentially like if the sandbox name kind of tells some part of that. It's like a place to try out new things. Some of them will become big, some of them will maybe a trickle down and die and whatnot. But there's a thing you need to get some scratchy things in delicate areas. Be careful. Yes, it's like a playground,
yeah exactly, so your experiment. Yeah, yeah, you can try out new things. It's more kind of you know, you have to do a more due diligence yourself if you want to take something into production. Likely, but you know there's a good stuff there. But incubating is a little more
sure. Then you get actual a little bit more because when you move between the levels, there's you know, security checks, architecture, there's requirements for having processes within the project so that you know every like scenario for example, if maintainers become unavailable and whatnot, like what's the contingency plans for everything, as well as like the development team's middle more maturity. Yeah I have more than one person accepting pull request, yes exactly. Yeah, I'm on vacation,
no coaching, yeah yeah, yeah exactly. And also there's like in that vein there's a like requirement for how many maintainers and whatnot or like different companies so that it's not like a one vendor only, yeah exactly. So then you move to the incubating phase, which is just like checks around these topics. So that's why you can know that, Okay, likely it's gone through security checks and whatnot. And then when you move to the graduated phase,
which is and I think there's an incubating level. Again, I don't remember exact numbers, but like, let's say twenty thirty usually or something max. It's much more smaller than the sandbox level. But then when you move to the graduated level a few years ago on then it was only cubaneites, for example, and they graduated level. Now we're talking maybe ten to fifteen twenty. Even if if there's been a lot of graduating greater cater graduated recently,
so there's nowadays. It's kind of picking of speed. But this is also a sign of communities maturing. Twenty five projects now and oh so many graduated and the ones you recognize harbor and hell yeah that's the big ones. Yeah yeah, so yeah exactly, So I need to catch up my numbers. I think I last checked them like half a year ago. So you know, it's moving quite fast nowadays, honestly. But it is. The scene is maturing, It is widespread adoption. So it just means that all
the tooling is really getting mature as well. So yeah, the graduated projects are your lowest risk. Yeah exactly, this is mature software. Go ahead and use it. I mean, I don't know that'd be all that shy about incubated software either, Like there's already people out there working on it. Like that's for sure, that's pretty interesting. It's nice to see that the scene and CF has gotten so sophisticated in how to manage an ecosystem, how
to grow it and move it forward. So the places people have a way to have confidence easy without being experts exactly. And I think it's really good because for example, for Keta, I don't know, like remember the except timeline, but it took them a while. Like if you go from incubating the graduate it might take like half a year to a year or or something because it is extensive, like you need to go through the checks and you need to pass and you need to really kind of get through it all.
So if you if you are looking at projects and they are at the graduated level and say here they like you know, you can trust them. You know that things are being thought off and whatnot, and like communities was the first one to go to graduated and think there's you know, things like from ethis Opah like whatnots at Helm Shelly as well, and well I have another question, but I think it's about time for a break, so we'll be right back after these very important messages. Hey Carl, here we have some
news from our sponsor, text Control. They just released version thirty two. Can you believe it? Of their document processing library, which includes new core functionality like document footnotes, SVG export, and much more in a great document edit, signing, collaboration and PDF processing into your asp net, Core and Angular applications With tx text Control, powerful libraries let your developer teams focus on
their core competencies while text Control handles your digital document processing. Check out all the new features and see the technologies in action by visiting the live demo at demos dot textcontrol dot com. And we're back. We're in Denmark with Annie Talvasto and some ping pong going on in the background. I'm listening to Dot rocks on, Carl Franklin, A Richard cavill and So I have a question
for you. So you're with a customer and they're interested in Kubernetes, and you know, obviously the first thing you do is you find out if it's a good fit for what they want to do. So what are the typical you know, what's your benchmark, what's your like? You know, what is it that you want to do? Like how big do you need to
scale? What are the questions that you ask to determine whether Kuberneti, kubernetes and containers in general might be overkilled for this company, that's a good question usually, so there might be a good number out there, but again it depends. I'm getting back to that all the time. But also usually if
it's at a certain scale, then it's good to go. What if they say, we have one hundred customers now, but we're going to have ten million customers next year by our you know, analysis or whatever, and you
know, do you what do you say? Then? Well, if they want to invest the resources and they want to go for it, I don't see an issue with it either, Like if they want to and if they're you know, I don't know, profit calculations for managing their infrastructure, tea team and the time they can spend on it and productivity and whatnot all matches. I think it's fine. And to be honest, there's like there's levels to it. And I don't do customer facing work on a daily basis nowadays,
so I don't know. I can't give you a customer example out of the top of my head right now, but I did recently speak with the startup who was about twenty people, thirty people, still so relatively stealth, so very early on, even in the cryptocurrency field, so very early on whatnot. And they actually had so I didn't speak with the CTO themselves.
It was just like, actually I think it was but anyhow, he didn't have all the answers to my questions because I got very interested because he said they're running playing cubunities and I was like, that's funky, like why would you do that in this day and age? But for them, apparently it had provided a really good set of environments and tooling because they wanted to stay as like bend Cloud and vend are Lucky neutral as possible, and they wanted
to do like switching between the clouds and whatnot. And I was kind of thinking, this sounds interesting. But if it works for them, hey works for them. You know, I've talked to lots of folks who think they want to be cloud independent but never actually do it, and even when they try, it's it's way harder than you think it's hard, and it's expensive and value. Yeah, you're trying to predict the future that may or may
not. My automatical reaction when you say, oh, we went with bare metal Kubernetes is so you had a person who knew their way around that or wanted to know their way around it, and that's the real reason you went that direction. Like in the enda, I think it does come down to individuals and what they want to work in. I don't know that why anybody picks a container strategy up front. It's more of a we've gotten the shape
of the software. Now we're looking at the implementation. And this is really an implementation detail like to be elastic, and it's really to control cost that I'm only when I'm only going to scale when it's going to make me more money, and I'm going to wind it down when I'm not, which pretty much means I have a cloud vendor. Yeah, because if I own the hard work, it's already paid for. What am I caring about the additional
electricity consumption of a deal? Yeah? Yeah, yeah. So we talked about this lately on dot net Rocks with various other guests about you know, over architecture. You know, if a company is just starting out and they have delusions of grandeur or or maybe wishes or desires, you know, in the perfect world, we're going to scale so big, so we better start
right now with you know, with with this thing. But if you build your applications in such a way that things are separated anyway, you know, whether it's even by DLL or by service or by just sections of your application that you can carve off when you need to, you know, maybe that's the time to think about Oh yeah, oh definitely. And I think it's honestly, I'm here saying, oh, go for it if you want,
and you have to watchet and whatnot. But truly, I think then one of the number one things to keep in mind is not to over engineer, not to overextend, to keep things as simple as possible, for like, well, not as long as possible, because whenever you add a new add on, a new CNC of project and you whatever, you're adding another level of dependency. And then, particularly in these days when you really do have to consider security, that's always another thing to check up on, that's always
another thing to think about. So I would advise always to stay cautious in that way that like, actually have a thing before you add on something. And I think that's a big part of it. And so I go around conferences talking about like a lot of different CNC of projects. So I'm the person who's like, ooh, kit is falco that hell this, try out everything, but truly don't do all the things, like do the things that matter to you, what makes sense for you. And architecture is the place
to start, right. You know, you keep everything in silos and loosely coupled, then it's easy to chip things away and move things apart exactly exactly yeah, and you don't, Yeah, then you can afford to be wrong too. We spend a week tinker with Kuberneties to try and deal with a couple of these issues, and you know, did work force are didn't it? It was another way to go about it. Can I peel some this code out running in lambda's or as vazure functions? And how do we feel
about that? Right? There's just a we all these discussions we've had about architectures, but really architectures should give you choice, yes, that flexible, yeah, and that you don't have to predict the future and you don't have to guess how things are going to go, that you can respond to your
business demand right and just move things as you need to move them. It doesn't We're not telling you, hey, you know you're you're delusional if you think you're going to be successful and not saying that, you're just saying that, you know, you don't have to spend the money up front if you don't need to, and if you are successful, great, If you architect things correctly, you'll be able to be able to switch when you need the scale exactly. And it is also about, you know, don't think about
tooling that you need. Think about what processes do you need or do you have already, and what is your architecture and what is your application and everything, and then fit the tooling to that, not the other way around, right, which I think is such an easy thing to say and like it makes sense, but it is tendency. I think a technical feel to be like, oh, that's the cool new thing. I want that, let's do that whatever it is which is you know, I get that that's cool.
As I said, I'm the person who's excited about all the new things. I'm advocating for them. But it is also you know, you should start with your needs, your processes, your team. Sure. Absolutely, I do feel like there was a period here where people will kubern eating because it had a good name, like they got over excited with so we will cover it aading all things, and then got in there and went, oh,
this isn't a free exactly a bit too complex. And it's why I'm really starting to feel more and more as I'm talking to folks who are working with this stuff, It's like you'll know you'll need it when you need it, right when you have that problem of how do I scale this way? Why are we strangling, you know, struggling at this point? And you know, I'm watching them logs and we're getting more more transactions and they start to fall off and I turn up the knobs on the cloud and it ain't
handling, it's stuck. And now that's that architectural conversation. And do pull things around and be able to scale them differently exactly exactly? Are there other tool space we talked about about the package management side, you know, how
do you set up that elasticity? Well, like I always worry about the administrative overhead of these systems of automating the scale and being able to see we aren't impacting our customers badly, like that we can see that the system is scaling up and scaling down to maintain a level of performance for the users.
Well, there's a well there's a lot of tooling that you could use from from CNCA for example or whatnot, which is usually quite a lot of them are like you know, the fact that you use with communities, but there's a lot outside of the foundation as well, And obviously, like after you maybe have package management, you have to think about monitoring, right. So Prometheus for example, is classic there with like Rapana and whatnot, which is
a big one as well. And I think it's also actually really important to keep the classics in mind because as project mature, they might lose some of the like the cool factor in that way, but they still are the backbone of of all of these infrastructure teams and whatnot. I found that monitoring is where you want to innovate anyway, Like you know what you want from a monitor, show me the stuff that's important, right, yeah, don't you
know. Prometheus has good templates for pulling the information it's going to be important, and Grafana has excellent dashboards for surfacing that information where you could look at it right away and go are you pay attention to this or everything's fine? Exactly exactly, Yeah, It's very funny that just like where do you want to innovate? What do you what do you need fancy new software? Like is there a possibility that a dashboard is going to make a difference beyond just
having a dashboard? Yeah, exactly, And particularly in the open source community, always try to kind of sneak in and remind of like if you are, for example, happily using from Ethio's helm and whatnot, you're thinking, oh, it's such a big project, so it's all good, it's open
source, it's are running. But actually some of the big projects might be the ones that need maintainers or contributions as well, because as I said, like if it's no longer the like the new cool kit on the blog, you might actually need a bit like more attention and sometimes sometimes to get contributions because also there's a lot of users, there's a lot of organizations tapping into it, so you also have a lot of help, require support and whatnot
needed as well as new future requests and everything. Then actually contributing to the old projects is super crucial as well to keep them up and running and keeping them viral and healthy. They are important. Yeah, I mean another recurring conversation for us is healthy open source community. And I mean obviously the founder the Cloud Foundation has this financial model integrated into it. If you're in that, in that system of incubating and so forth, there is compensation in there.
Where are they getting their money from? Like, how is that working? I don't know how the foundation well, I don't have the full list of sources for their running. But truly I would say that likely membership fees are probably a big part of it. Yeah, a big chunk of it, because there is If you look at the CNCF page for members it's a long list. It is a healthy list of really big companies, and when you get to the big level of sponsorships, it's it's also going to cost
you, Yeah, exactly. It also makes sense if you're a bigger company and you have bigger pockets and you're actually using a chunk of this offare or that the foundation has, it makes sense to contribute back to it financially, well, financially as well as resource. Yeah, exactly. It's always good to contribute in some way. And I would hope that sort of midband companies to get into the habit of hey, you depend on the software support it
exactly exactly some useful way it is. I mean the real question is do you do you allocate dev resources or just send money? Both both are reasonable because I mean the upside of the allocated dev resources is I want this feature. Yeah, yeah, let's go build it and contribute to the project exactly. And the most recent conversations are like generally when a now todays talk with maintainers, actually the thing that they usually ask for likely is financial investment in
terms of our investments and dev resources. Right, so you know, if you are a big user of a project, hire someone to just contribute to the project for example. Interesting, right, yeah, because yeah, because then you are using financial resources to provide those crucial development hours, contributions, features and whatnot. Because also you are likely if you are a big user of the project, you're likely also going to be the one who's asking for
new features, pressing the edges exactly. So then you can actually be like, hey, I'll get this person working on this so that we can actually get these features as well. And that is that is an existing model that people use also within the cop nating ecosystem. Yeah, it's always money involved, exactly. You've got to pay your people one way or the other exactly whatever they may be working on, and it it's cheaper to pay someone to
contribute to a Kubernates tool than it is to write that tool yourself. Yeah, exact, indeed. Yeah, it's a good it's a good mix either way. It's fun talking with them, though. I know if you feel like a lot of people who do this and it's like you ask them like, oh, what do you do and they're like, oh, I work for this company, but actually I only do contributions to this open source project. So like you know, you pil can choose which one I like work
for per se quote unquote. But you know I have that problem. People say what do you do? I say, how much time do you have? You mean want to I do for money? Yeah? It's yeah, workforce versus work on right, right, both both are very reasonable. I just don't know what I want to be when I grow up. That's the problem. I start now, that's not what's that about. I'm not starting that now. That's not a good idea at all. That's ridiculous. We
haven't talked about security at all. I don't even know if I want to. Yeah, yeah, well anytime that you take a piece of your monolith and split it off into another service, you have to think about security. Yeah. Yeah, you may have maybe using the same authentication methods that you're using on your monolith, but oftentimes you're not. So yeah, And there's a lot of good tooling that you can snap from the cn CF landscape there
as well, like both not the new ones per se. It's always a bit of like is it a new one or not, because some people Cubernites is new and for some people you're know, not the Sandford project, like, oh it's old news, right, But there's Falco for example, which is real time threat detection that you can use as well, and whatnot. So there's a lot of things you can plug into from the I don't but again, as I stress, actually use the ones that make sense for you.
But within my session I will do like here in this event, for example, we'll do a bit of at all, go and what not. So I do and oftentimes the cloud platform, like if you're using functions, dictates what kind of yeah, mechanism you're going to use. But you know, of course, if you have your containers and use whatever you want exactly exactly another reason you might go that way, right, I have specific security constraints, and yeah, the automatic stuff is not going to do it for
you, so it just gives you more choice exactly. I'll include a link to foll code looks interesting. So we've talked about security enough, we can move on. You talk about security. We always look security. It's our favorite developers love security. Yeah, well, some constraints have fun. I mean, there's no way to work in containers where you don't have security rules already in place. Like there's just the name we talked about this whole shifting
left of the security side. It's like these are the tools that shift you left. You're using it no matter what, so you mightage to you know, get comfortable with it and you don't have to retrofit it either. Yeah. And actually recently heard like a good anecdote and I think it's probably very well known one for security and for example for policy management, which is you know, a great key bowner as well from CNCF for policy management and security
on that fund. So as you kind of hear like you don't have to think about security or policy management as you know the bombers who are going to you know, come and limit you. You can think about them as guardrails. That works, like if you're driving in a steep cliffy area, for example, they're the guardrails that actually stop you from going over the hill.
But I'm going over the cliff. So you can actually go faster if you have good guard rails that are well thought of in place, so that then you can you know, speed as much as well, not as much as you want, but like a bit faster because you don't have to worry about
things so much. Are you're driving by braille, just bounce along that guard exactly, But like you know, you can use them as the guidelines, the guard lines, so that you can actually kind of well, you know, you speak to the right thing, which is what I'm doing is keeping you from being the code that caused the breach. Yeah, exactly right.
That by incorporating this the whole time and checking it the entire time, you don't have the code vulnerability that would eventually turn into an exploit for the company. Yeah. Ye, anyway, I mean, things move, nothing's ever certain, but at least try to some degree sure. So what's next for you? What's in your inbox? Oh, there's so many things. So there's the fall conference season is speaking up right now. I'll be in KCD
Washington, DC in a few weeks. Just yes, today the cube Con Cloud Native Corn North America program for this year was released and it's going to be in Chicago, and I'll be speaking there as well. Very looking forward to that as well. And there's a bunch of other conferences that I'm doing. And in addition to you know, I have I should maybe do a few more episodes of my podcast. And I'm thinking about the podcast. Oh it's been Oh we started in twenty eighteen, I think, so it's been
a bit. We've now been like a bit of a hibernation for like a yearsh or so, but we should maybe kick it back on. We started the first season by having essentially like actually we did like scripted episodes into what are containers, how to choose, how to buy a cloud, and whatnot, so like very like actually not that interviews aren't educational, but like very
like very deep except foundational details. Everything scripted, so every single sentence provided something new to attendees, and we got so much good feedback from very like beginner level people. They were very happy with the first season. But then honestly it was a bit two time, confusing consuming because it took like thirty hours to make like you know, minutes of Contra exactly. So then the second season we did or the season something that we've done like as an interview
format similar to this podcast as well. But now I should maybe revive it. But I'm also thinking about other stuff as well. I should start kind of I'm just trying to get this my head around, this idea of you can hibernate a podcast. What would that be? Like, I can't even imagine. If it happens, it will happen very organically. You just notice, oh, we didn't do an episode for a year. We'll see what
should I do now that we have? We do not this one, but we do episodes when there's something to talk about exactly there is anything to talk about. Why I talk about it? Yeah, sure enough. But like as far as what I try, I'm trying to act to my inbox or like when I'm actively adding as I go forward, I'm thinking about things like maybe a book or or training series or content video creation or something. Because I've been doing now international conferences for years already, and I last year for
example, with thirty conferences per year, which is a lot. So I'm trying to do other stuff as well. Conferences are great, but I think I want to do something a bit more long form as well, because constant session is it's great, but it's quite short lived. It happens carving up your whole life into one hour second. Yeah, exactly, So that's next. That's great. Well, thanks, it's been great talking to you perfect. Thank you for having me absolutely, and we'll talk to you dear listeners
next time on dot net box. Dot net Rocks is brought to you by Franklin's Net and produced by Pop Studios, a full service audio, video and post production facility located physically in New London, Connecticut, and of course in the cloud online at pwop dot com. Visit our website at d O T N E t R O c k S dot com for RSS feeds, downloads, mobile apps, comments, and access to the full archives going back to show number one, recorded in September two thousand and two. And make sure
you check out our sponsors. They keep us in business. Now go write some code, see you next time. You got metavans dot