Sponsored by Reblaze, creators of Curiefense
Justin Dorfman | Richard Littauer
Hello and welcome to Committing to Cloud Native Podcast! It’s the podcast by Reblaze where we talk about open source maintainers, contributors, sustainers, and their experiences in the Cloud Native space. We are super happy to have two guests with us today, Zbynek Roubalik, a Senior Software Engineer at Red Hat, and Tom Kerkhove, Azure Architect at Codit. Also, Zbynek and Tom are both Maintainers for KEDA. They are here to tell us more about KEDA, why it’s so cool, what they do as Maintainers for it, their challenges, and personal goals they have with helping them become better Cloud Native Engineers. There is a discussion about how it’s not just about code, but all the other pieces that really make a project, and knowing how to manage everything in between. Go ahead and download this episode to hear much more!
[00:02:00] We start with finding out what KEDA is, how long it’s been around, and how large the open source community is.
[00:04:20] Justin wonders if Tom and Zbynek were there from the beginning of the sandbox onboarding or if they came after.
[00:06:32] Tom tells us about doing case studies they’ve done for Alibaba Cloud and the CNCF blog.
[00:08:35] Richard asks Tom and Zbynek why this project and what is so cool about KEDA compared to anything else in the Cloud Native space, and they tell us what sort of stuff they do as Maintainers for KEDA.
[00:11:47] Justin brings up that he saw “Merch” in their store, and he wonders how Tom and Zbynek are fulfilling that and if they use a service.
[00:15:36] Richard wonders about Tom and Zbynek getting paid to do this work, and Tom fills us in.
[00:16:33] Tom and Zbynek talk about the major challenges that they’re facing, which include “commit and run.”
[00:24:00] Richard asks Tom and Zbynek to share their personal goals for being with this project, where do they want to be in five years, and how is KEDA helping you become a better Cloud Native Engineer.
[00:27:52] Richard mentions Tom’s website you should check out and he asks him if people could only read one of his blog posts which one would it be.
[00:30:28] The guys all discuss how it’s not just about the code, but about all those other pieces that really make the project and knowing how to manage everything in between.
[00:33:09] Find out where you can follow Zbynek and Tom online.
[00:33:45] We end with Justin asking Tom and Zbynek who they would like to thank at the CNCF or anyone in the Cloud Native community.
[00:02:05] It’s basically aiming to make application autoscaling that simple on Kubernetes because it isn’t. So, we try to make it super simple for you so you can focus on your application and not the scaling internals of Kubernetes.”
[00:05:07] “Funnily enough, that’s a very hard piece of maintaining open source, knowing who is using it because you cannot measure it in any way.”
[00:24:29] “By not having to worry about how Kubernetes scales.”
[00:24:37] “I bluntly set this one. We would present the KEDA for the incubation graduation. I want this to be the standard application autoscaler so that everybody uses the same thing, that you don’t have to worry about it, and we just handle everything for you, frankly.”
[00:24:46] “I like the simplicity because I like when things are from the user perspective because usually, we are developers, engineers, and we are focusing just on the codes, technical side.”
[00:25:49] “And now I’m sometimes exaggerating, but they still understand how to just scale this freaking thing, while it’s just the helm install away and you’re good to go.”
[00:26:30] “That’s also a good example because we are working on a new reference case with CAST AI and they basically use KEDA to make Kubernetes cost efficient and make sure that we don’t waste electricity and all this kind of stuff and make the environment a better place just by optimizing your workloads.”
[00:28:58] “And that’s why for my projects I always use the mantra which is, It’s just a pull request away.”
[00:29:48] “There’s so many things you need to do which is not code, it’s ridiculous.”
[00:30:54] “I haven’t written a single letter of code for KEDA. I’m doing all the other things.”
[00:31:41] “I would like to say that this is super important, like not just doing the code as mentioned, but all the stuff around, it is very important to have like active project and the health, because if you don’t have the healthy environment around, the project was stale basically.”
[00:31:56] “I fully agree on one the typical things that I see is you go to a project and you look for documentation. If you’re lucky you find documentation, then you ask, do you have any documentation.”
[00:34:09] “So, Chris Aniszczyk, the CTO of CNCF, because although he’s super busy, he’s always supporting us definitely. The fact that he’s helping KEDA from the early days when we were so small project and he believed in us from the start.”
- Executive Produced by Tzury Bar Yochay
- Produced by Justin Dorfman
- Edited by Paul M. Bahr at Peachtree Sound
- Show notes by DeAnn Bahr at Peachtree Sound
- Transcript by Layten Pryce
Tom [00:01]: We have some case studies on the blog, for example, Alibaba Cloud. We did one even for the CNCF blog, but I never taught I would say this about myself, but I'm really one of the PMs for this project. And I don't mean this badly to a PM, but I never saw myself as a PM, but that's just all I do reach out to people. Hey, cool stuff. Can we do something together? Because that really helps people convince that the project is used. People rely on them. Certainly, if you see big names, if they use it, then we probably should use it as well. It must be good. And that really helps. In terms of the CNCF incubation, luckily there end users can talk off the record. So we have some of the people with the big companies that we cannot list that are apart of it. Still you want to talk to your friends and colleagues about the ones. And obviously there are still ones that we do not know about because it's free in open-source
Richard [01:00]: Hello and welcome to committing to cloud-native, the podcast where we talk about the confluence of cloud-native and open source. Super excited to talk about our guests today and to talk with them even cause that's what we do here on the Committed To Cloud Native podcast, besides me, Richard Littauer. We also have Justin Dorfman as a panelist. Justin, how you doing?
Justin [01:20]: I'm doing great. How are you?
Richard [01:22]: I'm doing excellent. Happy to be here talking with two guests today. One of them is calling from my favorite city in the Czech Republic from Bruno and we have Zbynek [name], which I probably totally butchered his name for which I apologize sincerely. I do not speak Czech. I am so sorry. We also have Tom Kerkhove. Zbynek is joining us today as a senior software engineer from Red Hat, whereas Tom is joining us as the Azure architect at Coded. Both of them together are maintainers for KEDA, capital letters please. What is KEDA? Which one of you wants to go first?
Tom [02:05]: Well, I can get started. It's basically aiming to make application auto-scaling that simple on Kubernetes because it isn't. So we try to make it super simple for you. So you can focus on your application and auto scaling internals of Kubernetes.
Richard [02:22]: I thought Kubernetes wasn't natively made to scale. Isn't that the entire point of like Kubernetes? It's made to help make it easier for you to do that?
Tom [02:29]: Definitely. If you went through the whole internals and the guts of Kubernetes, and you understand how to get your external metrics inside the cluster and all these kind of things. Like if you just want to scale CPU and memory fine, you're good to go. But as soon as you want to get it from, let's say Prometheus's or some cloud provider, you need to do some more work and we try to give you a consistent approach to do it for all your applications, be it with cloud vendor, be it with some open source technology, we give you a consistent experience.
Richard [03:02]: Okay. That makes a lot more sense. How long has this project been around?
Tom [03:06]: It really was created in 2019. So I just kind of go by that.
Zbynek [03:09]: I would say it's like [03:11 inaudible]
Richard [03:12]: How large is the open source community? How many developers you have working there and how many maintainers?
Tom [03:19]: We have four maintainers, so two people from Microsoft and us too. And I think, how many contributors do we have?
Zbynek [03:27]: More than hundreds.
Tom [03:29]: It originally got announced in May, 2019, indeed. So as a partnership between Microsoft and Red Hats to close the gap and given event driven, autoscale it and then early last year, so that's 2020, we donated it to the CNCF as a sandbox project to basically stimulate the vendor naturalness and then actually now we are working on our proposal for incubation and we're actively doing end user interviews. So it is a great time to have this conversation because we are also learning from people how we are helping them make that auto-scaling simpler.
Justin [04:08]: That's awesome. I mean, the reason I reached out is because you're a little ahead of us in the program. We just joined sandbox and I noticed that you're getting ready to go to incubation. Were you there from the beginning of the sandbox onboarding or did you come after?
Tom [04:26]: I jumped in around sandbox, I think Zbynek, you're part of KEDA from the start.
Zbynek [04:32]: I joined couple months later, so maybe four months later in the summer of 2019.
Tom [04:40]: So it's mainly presenting what the project is, how it helps people and why it should be part of the CNCF. We came at the moment where they were just changing the whole process. So it was a bit finding out what the new approach is, but I think it was fairly straightforward to get it done. And that also helps you think about how you run the project, helps you to think about security, being a good member of the community. I know now with incubation, more about what's the adoption, what's the community and who are the end users. Funnily enough, that's a very hard piece of maintaining open source, knowing who is using it because you cannot measure it in any way. So you really need to reach out to people. I noticed you use it. Can we list you as an end-user? And I was actually very surprised how many times you hear, okay, I need to check with legal if we can do it, this and that.
Justin [05:31]: Yeah. That's so true.
Tom [05:34]: Because you're only using free stuff. So for me, it seems obvious to just give back the logo, let's say, but of course a logo has.
Justin [05:43]: They are strict. We had eBay's logo up on our site and they're like, we didn't say you could use that. We're like, oh, okay. Our bad we'll take it down. Like, it wasn't a big deal, but it was just, they track it down and they're like, oh, we can't have that onsite.
Tom [06:00]: Like we have some really super cool end users. We already have some listed publicly, but we have some other great ones, but we cannot talk about them. But man, they're so cool that it's really annoying because people use it from day to day.
Justin [06:16]: Yeah. Then the most annoying thing is that in order to go to incubation from the sandbox is you have to have user studies. And you know, we can't write them because they won't let us, it's like this kind of catch 22. So I totally feel ya. I mean, you've got some really big brand names. You've got Microsoft, Alibaba Cloud and a few others, but are they doing the case studies? How do you go about doing that?
Tom [06:41]: We have some case studies on the blog that we have, for example, Alibaba Cloud. We did one even for the CNCF blog, but I never thought I would say this about myself, but I'm really one of the PMs for this project. And I don't mean this badly to a PM, but I never saw myself as a PM, but that's just all I do reach out to people. Hey, cool stuff, can we do something together because that really helps people convince that the project is used. People rely on them. Certainly, if you see big names, if they use it, then we probably should use it as well. It must be good. Right? And that really helps. In terms of the CNCF incubation luckily there end users can talk off the record. So we have some of the people with the big companies that we cannot list that are part of it. So that helps. Still, you want to talk to your friends and colleagues about ones. And obviously there are still ones that we do not know about because it's free in open source.
Justin [07:40]: We recently had a guest on Avi Press co-founded and CEO of scarf.sh, check that out. Him and his team are figuring out the problem to see who's using your stuff. It's pretty cool. We're going to implement it pretty soon. So this is cool. Like we can share, you know, tips and stuff. I like that.
Richard [08:01]: Also a [08:01 inaudible] KEDA is for those of you who want to go to the URL, Tom, I have a question for you. So you've been working for Coded as an Azure architect for a while, but you're also speaking of big names. You have a lot of like, you know, medallions to attach to yours. So you've been a marketer of Azure and MVP and advisors since 2014. You're one of the first GitHub stars. You've been a CNCF ambassador since 2020. And you do a lot of different stuff around GitHub, [08:25 inaudible] Promitory, kubernetes event, grid bridge, Azure deprecation. Arcos, I'm literally just reading your bio, but partially because the people in the audience haven't read your bio and it's my job to make sure they know who they're listening to. So one of the questions I have for you is why this project out of other projects, what's so cool about KEDA compared to anything else in the cloud-native space.
Tom [08:45]: Honestly, I have a customer who went from an Azure service to Kubernetes. They said, we'll use Kubernetes. I think that was four or five years ago. It's great. It's contained. And it really scales very well. And then they deployed it and then they said, okay, let's auto scale. And then I was like, this is not supported out of box, good luck. Back then, there were only two metric adapters, which was Prometheus and I think GCP and I was like, okay guys, you can build it yourself or you manually scale? They said, well, manually scale. But that's how I started [09:19 inaudible] to bring Azure monitor metrics and Prometheus.
[09:23] Fast forward to the partnership with Red Hat and Microsoft, I was like, man, this is what my customer needed. This is just closing such a big gap. And actually I was already working with Jeff Holland before as an MVP. So I reached out and say, Hey, how can I help? Because this impacts all my customers and yeah, I really believed in the project, I started contributing and then the team asked if I wanted to become a maintainer. So that's the story. It's just what I needed, basically.
Richard [09:54]: Awesome. Zbynek what about you?
Zbynek [09:55]: Oh, it was just like totally different because as I mentioned, the project originally started as a collaboration between Microsoft [10:02 inaudible]. So there was the [10:04 inaudible] the guy from [10:06 inaudible] that told me, okay, we are working on this project and we are looking for some additional workforce. So I was looking at it and I was like, this is cool because this brings something new to Kubernetes, something that's missing because all the scaling is not simple to mention. So that's why I joined this project because I like the idea.
Richard [10:27]: So you two are maintainers, but maintainers as a term, which covers all manner of sins. What do you say that you're jobs are on a day-to-day basis. Are you merging issues? Are you doing big strategy planning? Are you community facilitators? What sort of stuff would you say you do?
Zbynek [10:42]: I wouldn't say that we are doing everything that you mentioned. So basically we started with the communication, we have issues, we have discussions, on slack we have [10:50 inaudible] on Kubernetes space. And we are making decisions, [10:57] like feature, features that they would like to have, would like to implement and all this stuff. And it's complex so you can try different aspects of project development as I mentioned, slowly transferred into [11:10] because [11:11] well this kind of [11:12]. So that's just perfect.
Justin [11:18]: No, that's how I know of Tom is Twitter. And you guys, especially, you are like besties on Twitter. Like it's this constant quote retweets and hearts and it's really cute. I just want to go into something. And the other thing with the CNCF is obviously built in the community and that's something that I am, where it's like I have multiple hats. The biggest thing is great in the community. And I saw you have merch in your store, like, how are you fulfilling that? Do you use a service or what do you do?
Tom [11:57]: So that's a good example of the benefits of CNCF, where they have a service desk. And I don't know how it originally arose, but we thought, yeah, why don't we have Merch? So he started digging, and there's a lot of options. There's like these websites where they can upload your logo and then people can buy merchandising and they just pay for everything I think. But I just said, Hey, let's go to the CNCF service desk and see what the guidance is because we cannot be the only ones with this request. And then they told me that CNCF projects, which are part of graduation or incubation, get this for free. So they become part of the CNCF store. So I said, Hey, no problem. I fully respect that. Do you have any recommendations? And then they just ask, okay, what's the quantity? And we said, yeah, we just want to do a limited amount. And then they said, okay, let's just do it for you. So they picked it up for us. So that's a good example of how a foundation helps.
[12:55]Also in terms of the websites I think it was with the 1.0 release. I created the first ugly version, just to have something up. And then we involved CNCF as well. And they had a person come up with the kid and design and built it for us. So they really help with these typical non-project specific things, which save us a lot of time.
Justin [13:17]: The docs are beautiful. They just flow really well. They're really well done. I was pretty impressed with that. So it's interesting. So you're a sandbox project, but you're kind of getting incubator slash graduation perks. Am I going to have to talk to Chris [Name]about this? What's going on?
Tom [13:36]: I always say one can only ask, if it's a no, it's a not. Yes, it's even better.
Justin [13:43]: Yeah, you're absolutely right. And that's a good lesson for everyone, is just, all they can do is say no, there's really nothing to lose. Interesting. Okay. That's good to know. So you don't have to do any of the fulfillment, they handle everything. The Shopify store, I believe is what you use.
Tom [13:59]: Yeah, I think so. But they did everything for us indeed, but there are actually a lot of ways where you can offer this. And so I was actually a bit overwhelmed and they wanted to use what is typically used. So we got involved.
Justin [14:12]: We use Real thread, they do a really great job. We could provide them, they make a Google form and people fill it out and they put t-shirts stickers. Anything you want to put in there and they ship it anywhere in the world. So they've been a really great partner and yeah, we just basically get our stickers from sticker mule, ship them to Realthread. And then they just handle everything. It's a really great service.
Richard [14:37]:Justin used to work at sticker mule. So just take whatever he says with a grain of salt.
Justin [14:42]: I think sticker mule makes the best stickers like really do, but what do I know.
Tom [14:47]: But you still need money to be able to make stickers. We don't have any funding so, well at least to my knowledge. we really rely on the good people, such as Snake, Azure dev, GitHub. I don't know [15:01 inaudible] we have.
Zbynek [15:04]: It's [15:04 inaudible] for hosting our [15:06] for testing.
Justin [15:08]: Well, are you allowed to like, do GitHub sponsors or does the CNCF prohibit that?
Tom [15:15]: That's actually a good question.
Justin [15:16]: Cause like, if you did like a GitHub sponsor, you can put it into an open collective and then get like a virtual card and then pay for stickers out of that or just pay for them yourself and then get a reimbursement.
Tom [15:26]: Yeah. That's a good question. I don't know if we are allowed to.
Justin [15:30]: Got to have stickers or it's not an open source project.
Richard [15:35]: So you two are paid. You have money coming in. You're not doing this of your own initiative. It looks like you're in nice apartments. So I'm just making sure, I mean not too nice, but nice.
Tom [15:44]: I do this in my spare time. So that's also why I'm super slow and delaying everything. So my company gives me some budget too, as an MVP to do whatever I want. So that helps. But the majority is while my kid is sleeping or when my wife is out for work.
Justin [16:05]: How many times has Microsoft tried to recruit you, Tom? It seems like you are one of their biggest advocates.
Richard [16:14]: Well, Justin is trying to get a referral fee there again, sneaky guy. You got to watch that dude. You've got to watch him. So funding aside, CNCF has given you a lot of opportunities and obviously, you know, your companies allow you to do work on this stuff. That's great. It seems like a sustainable project. You have awesome docs. You have tons of contributors. What are the major challenges that you're facing? Like what are like the top two or three things that come to mind where it's like, oh, this is the problem that I have, and I don't know how to solve it?
Tom [16:43]: Commit and run. So we allow you to contribute scalers. And while it's super simple to add them, it's also, I think the downsides, because we have scalers to systems that we're not experienced with. I'm looking to see if I have an example.
Richard [17:01]: So on your homepage, you have this huge list of projects. Those are the scalers you're talking about, right?
Tom [17:06]: Yeah, so we rely on the community. For example, we have one for who are way clouds. If there are issues. I mean, nobody used that before. So that's why Zbynek now said okay every new scaler needs to have those end to end tests, if they're not there, sorry, we will not be able to merge them because we have no idea how it should be.
Richard [17:26]: You must be this tested to ride the ride, is what you're saying.
Tom [17:30]: I think that would be my remark.
Zbynek [17:34]: I agree this is the biggest issue we have because as you can see [17:39] scalers is huge and it is not in one person. So two persons to do all these kind of stuff. So we need to keep the people engaged that are using the specific scale. So usually [17:55] is that there is somebody that, okay, I would like to have this scaler for KEDA what can we do? So we encourage them to continue [18:01] and would really appreciate it if thIS person can take a little kind of [18:06] and from time to time [18:09].
Richard [18:11]: I wonder if maybe a document policy talking about scalar maintenance and what's sort of required of you, if you become someone who submits a scalar might be a good idea, because then you can say, if you aren't reactive to issues, we'll remove the scaler from production because it's totally going to.
Zbynek [18:23]: If I'm not mistaken, we have the [18:24] kind of [18:27] documentation software. So once those scalers are implemented [18:32 ] . So that it's fine.
Richard [18:35]: How many users would you say you have?
Tom [18:37]: It's open source, we have no idea.
Richard [18:39]: No idea at all. Okay. But tens of thousands, hundreds of thousands of people using it. No idea. Super cool.
Tom [18:45]: We only know about the people that are listed at least what I know. I don't know about you Zbynek, but that's why they becoming a maintainer makes you see these kinds of things. But that's why, if you use an open source project, it's the least you can do.
Zbynek [19:02]: I know that Jeff mentioned the number of installations of Kubernetes but I'm not sure if I'm allowed to talk about this. [19:12 inaudible] was pretty impressive I would say for this kind of small project.
Richard [19:18]: What's that thing about Huawei cloud. I don't know, I hadn't heard of all Huawei cloud before. I presume I will hear from it again, but I have no idea what Huawei's user base would be. I don't speak Chinese. I can't read Chinese documentation and I don't know a lot of Chinese developers. And so I was like, it could be freaking massive and maybe there really is one person keeping all that up and you wouldn't know, super tough problem.
Tom [19:42]: For example the Alibaba Clouds, they are using it for, like scaling the applications that are being deployed on the Cloud. So I would say this cloud is pretty huge, like in China, so we like [19:51] number of customer that are being like the customers of Alibaba service, is just one customer for us.
Justin [19:57]: We had William Morgan from Linkerd on and one thing that he does is he has an adopters.md. I don't know if you have that, but he basically just goes out and anyone he finds out that is using the project. It's not about so much the logo on the site, but the easy way for someone who's using it can do a pull request and then add their company. That's kind of something he's very obsessive about.
Tom [20:25]: Yeah, I think that's another good approach, but I'm just thinking, what's the difference. Of course it's still a logo, but I think we would still have the bigger company saying it's not okay that you list us. It will be easier of course, but I think the way this needs to be fixed is not on our project. But for example, if you have a look at GitHub, you have the star. Star says nothing. I mean, this is just reds bookmark is repo. But what if there would be something new, like a button. I use this project and then you as an individual could just click that button. I could click it and then you at least have a better measurement.
Richard [21:06]: There's no way that's not coming. There's no way that Microsoft doesn't know that's something that should be implemented for GitHub. I mean that's been obvious for years.
Tom [21:13]: I've suggested it but let's say that it's not getting much traction.
Justin [21:17]: Maybe we have the power Richard to make this happen because we know some people.
Richard [21:24]: We can figure it out. I'm wondering if it's possible. I used to have the New York times logo on my personal website because I was [21:32 crosstalk] random like throwaway article about dothraki the language in the movie Game of Thrones and then New York times emailed me saying, you're using our logo on your website in a way to advertise. You are doing it to increase your cloud. That will cost you $3,000 a month or something. And I'm like, ah, no, thanks. So it was something, I don't even remember. I just took it down. Like, my commit on GitHub was like, these guys are jerks, but I realize now what I could have done is made a section of the website saying I can neither confirm or deny these people, let me use their logo or not.
Tom [22:09]: But how do they know so quickly?
Richard [22:11]: No they didn't know very quickly. It took them years to find out, but they basically have armies of minions who go out and just try to make money. I mean that person's making his pay grade and you know, I could have been nicer. I wasn't, but that's cool. But like that's just someone's job.
Justin [22:26]: They have like the same type of job of someone giving out court orders to people.
Richard [22:31]: Just like a patent troll almost, you know, numbers.
Justin [22:38]: You A hole.
Zbynek [22:40]: Regarding the patent trolls this is one thing I like about [22:43] is i can say the same, but basically [22:46] what is reasonable and it's been like developed by some [22:52] he can file for the patent and it's like some report from the company, but the company will take over the patent and will put it outside for everyone to submit [23:03]. So you can work on the patent, you get paid if your name is still on the patent [23:08] the company [23:10] and there is like [23:10] it will be like free forever basically.
Justin [23:14]: I'm not sure if, how would IBM feel about that? Because that is the parent company of Red Hat.
Tom [23:21]: Well we are still different company. So they are just owning the shares. So basically we have different policies. And this is still [23:30]. I know that in the agreement when IBM was buying Red Hat, there was like this note, just the patent will remain [23:37] as promised, this is good.
Richard [23:46]: So you can get different perspectives. This is a question I haven't asked a lot, but you two seem super personable. And it's an interesting podcast just with maintainers, which we often don't have. Sometimes we have people who are more like sales reps or advocates or evangelists, my least favorite word in the world. What are your personal goals for being with this project? Where do you want to be in five years? How is KEDA helping you get there? Specifically in relation to cloud native? Not like I want a house somewhere in the mountains of Poland, but like specifically like, what is KEDA helping you do? There's a couple, there are small Hills. They are not very big. Okay. I'm sorry. I'm sorry. The mountains that are [24:23] Tom, where you live. I'm just curious, like how is KEDA helping you become a better cloud native engineer?
Tom [24:28]: By not having to worry about how Kubernetes scales. I bluntly set this one, we would present KEDA for the incubation graduation. I want this to be the standard application autoscaler so that everybody uses the same thing. They don't have to worry about it. And we just handle everything for you, frankly.
Richard [24:48] Nice.
Zbynek [24:48]: I would say the same, I like the simplicity because I like the way things from the user perspective because usually we have the [24:57] engineers and we have [24:59] technical side [25:03]. So I do like that it should be as simple as possible even for people who don't know Kubernetes, don't know containers that much and don't know all the internals. They just have a application [25:16] they just want to scan it. So this is the end goal I would say, it shouldn't be very simple.
Tom [25:23]: I fully agree. And then I typically hear a lot of people say, Hey, but you don't need it is because you could do it this way or that way. And I'm like, that's fine. But you understand Kubernetes, your highly involved in the community. So you know all the projects, but imagine you're a person who's new to cloud native and he's responsible to deploy his application. Why should he spend days, sometimes exaggerating, but days to understand how to just scale this freaking thing while it's just the [25:55 home install] away and you're good to go. And that's one less thing to worry about. There is already so many things at Kubernetes you have to worry about.
Zbynek [26:06]: Kubernetes is there [26:07] for [26:07] little things, so there is like this [26:09] just called horizontal auto scaler, which allows you to scale your application based on the CPU and [26:17]. There is one big difference because this component doesn't allow you to scale to zero and KEDA can scale to zero. So you can scale your application down to zero. The [26:26] can just scale to one. This is the big difference I would say from a typical perspective.
Tom [26:33]: Yeah and that's also a good example, because we are working on the reference case with cost at AI and they basically use KEDA to make kubernetes cost efficient and make sure that we don't waste electricity and all this kind of stuff and make the environment a better place just by optimizing your workloads. And that's a nice reference case because that way we're also helping the environment.
Justin [26:59]: Elon Musk would love that. And as you said, it would just be a great marketing angle. Like, you know, good thing, you know, you're not flexing. You're just saying, Hey, look, we're saving energy. Not only are you saving money, but you're saving the planet.
Richard [27:14]: For other listeners who may not know, Justin and I are both panelists on another podcast called Sustain. We're literally talking about sustainability of open-source and every now and then it goes away from maintainers, stopping burning out to sustaining the environment. What I love about this podcast is that it basically could go on either one of Justin's and I podcasts. It's basically the same. It's like, how to sustain maintainers and the environment super cool. It makes me really happy.
Justin [27:39]: It's like a parallel universe in a way.
Richard [27:41]: Crossover.
Justin [27:43]: Yeah crossover. What is this a crossover episode? Anyone who watches BoJack would understand that.
Richard [27:50]: So Tom, I'm looking at your website right now for listeners. This is blog. Tom Kerkhove, that's K E R K H O V E .BE. And it seems like you're a prolific writer. What blog posts do you have down the line? What are you most excited about, telling people, if people could only read one blog post, what would it be? Again really open-ended question cause mostly I just love hearing your responses.
Tom [28:14]: It would be one that's on my to-do for almost one to two years, which is about my adventures of doing open source as a maintainer.
Richard [28:24]: Can you talk a bit about what's on that?
Tom [28:26]: First I need to have some time to write. It's basically going to talk about what I learned from being on the other side of the fence. And one of the biggest lessons to me was instead of writing a gibberish issue, with a vague bug report, really take the time because it will also help you get the bug fixed faster. But instead of opening a bug, sometimes it's just as simple as opening a pull request. And that's why for my projects, I always use the mantra, which is, it's just a pull request away. So that's why documentation, automation, quotes or whatever has to be in the git repo. So you can just send the pull request and that's why CI/CD and all of those things, which is now all in, and [29:17] is such a great help because you don't have to rely on the human making a manual change before the CI succeeds, for example. But yeah, I need to write about that first.
Richard [29:27]: No, thank you for sharing. That's awesome. I can't emphasize enough how important it is for people to know that in writing open source, isn't about your intent and it's not just about just showing in coding, but it's about actually thinking about how to be effective and how to be effective at being a nice person.
Tom [29:47]: There are so many things you need to do, which is not code, it's ridiculous. Even if it's replying to issues and discussions.
Zbynek [29:58]: Those people should sometimes bare in mind that basically we are spread across the globe. So there are different time zones, certain days of week. This is another aspect of being maintainer. To respond toward report, at least I try to respond.
Justin [30:13]: Richard's working on a code of conduct issue on a project that we work on.
Richard [30:18]: No it's fine,it's so draining, man.
Justin [30:22]: That's the thing it comes with the territory. Like, as you said, Tom, it's not just about the code. It's about all those other pieces that really make the project. And if you want to be going from the sandbox to graduation, those are the things you're going to have to deal with. So anyone listening to this, they're thinking about bringing their project to the cloud native, to the CNCF. These are things you have to think about. It's not just writing the best code. It's how do you manage everything in between?
Tom [30:52]: I haven't written a single letter of code for KEDA. I'm doing all the other things. It's not nice, but.
Richard [31:04]: No that's just so great to admit, like that's exactly what maintainers do, is not write code.
Justin [31:11]: The only code I write on the main Curifense project is the MD, you know, the markdown files, updating the read MEs and getting the release notes together. And that's fine because that's what I'm good at. I'm good at doing the other things that are not co-related. So building community from scratch, very difficult, but talking to you and everyone else that also has CNCF projects, it makes it a lot easier. So thank you both.
Zbynek [31:39]: I would like to say this is super important, like not just doing code as Tom mentioned but all the stuff around it is very important. Like [31:48] project and healthy because if you don't have the healthy environment, the project will [31:52].
Tom [31:55]: I fully agree. One of the typical things that I see is you go to a project and you look for documentation, and if you're lucky, you find documentation, then you ask, okay, do you have any documentation? Yes, it's the code? And I'm like, man, I'm not going to go through the code to understand how to use it. It should be documented. Otherwise I'm out. I mean, I'm here to use your free stuff, man. I don't want to waste my time. I'm just kidding, but you really need to think from an end user perspective, otherwise you will not build a good product in my opinion.
Richard [32:29]: I like what Zbynek was saying earlier about UX or the DX, as they're sometimes calling it the developer experience super important work.
Zbynek [32:36]: Let's say you would like to try some new project and they [32:38] documentations, [32:41] it fails and you don't know what to do because the documentation is not very good. So this is something that is very important for projects. Even like [32:52] projects but all the projects that are [32:54].
Richard [32:55]: Every listener who has had this experience, if you're not driving, raise your hands. Yes, that's right. That was all of us. Cool. Thank you so much. That was really awesome to get your insight into being a maintainer for KEDA, to being a maintainer in general, it's being good people. Where can people follow you both on the web? Zbynek where can people follow you?
Zbynek [33:15]: I have a Twitter account, which is ZROWALEK, which is pretty hard so you're going to find it somewhere. Yeah this is the main point for reaching out to me.
Tom [33:31]: I fell like drinking from a fire hose. You can follow me on Twitter. For links of the show notes.
Justin [33:41]: Before we close out, I want to do maybe like, maybe this is like a new segment, but basically who would you like to thank at the CNCF or anyone in the cloud native community? I think that's important because then people can follow them and make sure that they're getting more points of view. So Tom, who would you like to thank?
Tom [34:03]: I would say Liz Rize because she's helping us with the incubation. She's really nice to work with. But also Chris, so Chris, the CTO of CNCF because although he is super busy he's always supporting us. Definitely the fact that he's helping KEDA from the early days when we were so small project and he believed in us from the start. He's really nice, given he's that busy. And he's also the one that said to me guys, why are they not incubation yet? You need to start moving. So that's a true sign that he believes in what we're doing. So I think he's a really nice person.
Zbynek [34:44]: Inaudible
Justin [34:45]:You're for Chris as well. Yeah, I agree. He's always on a service desk. I'm like, how do you find time to do that?
Tom [34:53]: I don't know if he ever sleeps.
Justin [34:55]: I don't think so.
Zbynek [34:57]: Well Tom if you say it's [34:58] I'm not sure because you don't sleep as well because you're up doing all this stuff.
Justin [35:05]: Robot conspiracy,
Tom [35:08]: Maybe or it's just about priorities.
Richard [35:13]: Awesome. Thank you both so much. It was great having you on, let us know in the future if we can help out at all. Listeners, if you're curious about them, you can follow them in the show notes. If you have any comments about this episode, please send me or Justin a line. Justin@Curiefense.com. That's JDorfman@Curiefense.io. Again Tom, Zbynek [35:41] whichever you are. If you're in Brussels and a [inaudible]. Thank you so much.
Thank you. Thank you for having us.