1 00:00:07,440 --> 00:00:09,570 Tariq Islam: With an interest first, right? Yes. 2 00:00:12,450 --> 00:00:15,030 Jamie Duncan: Thanks for joining us joining us today on episode 3 00:00:15,030 --> 00:00:18,180 10 of the K files. My name is Jamie Duncan. I'm a customer 4 00:00:18,180 --> 00:00:22,830 engineer at Google, with me as always, is john Osborne. I run 5 00:00:22,830 --> 00:00:25,080 this as john Osborne, chief architect for Red Hat, public 6 00:00:25,080 --> 00:00:30,240 sector. And tark Islam. Everyone, this is Tarik Islam, 7 00:00:30,330 --> 00:00:34,380 sales engineering manager at Google. Um, we've been having a 8 00:00:34,380 --> 00:00:36,690 discussion for about 10 minutes. Now before we click the Record 9 00:00:36,690 --> 00:00:40,560 button. We know what we want to talk about, but we're not 100% 10 00:00:40,560 --> 00:00:41,910 sure what to call it. 11 00:00:43,020 --> 00:00:45,450 We've called it get ops and it's not quite get Ops, we've called 12 00:00:45,450 --> 00:00:49,830 it ci CD. And it's not quite that. JOHN, you said the CN CF 13 00:00:49,830 --> 00:00:52,950 has a name on application development and delivery, 14 00:00:52,950 --> 00:00:53,880 something like that. 15 00:00:54,270 --> 00:00:57,060 John Osborne: Yeah, so Helm is a is a CNC f project that 16 00:00:57,060 --> 00:01:01,710 graduated this year. And they have it filed under application 17 00:01:01,710 --> 00:01:05,760 definition. And deployment, I think, okay, 18 00:01:05,940 --> 00:01:08,790 Jamie Duncan: but that's your target quote was immediately 19 00:01:08,790 --> 00:01:11,730 hated it. Yes. Like 20 00:01:12,450 --> 00:01:13,980 Tariq Islam: that right under thanks. I hated 21 00:01:14,430 --> 00:01:16,050 John Osborne: definition and development. That's what they 22 00:01:16,050 --> 00:01:18,960 have it. Mission development, probably 23 00:01:18,960 --> 00:01:21,270 Jamie Duncan: the most accurate definition tark throughout was 24 00:01:21,270 --> 00:01:27,570 called up CD. So Ops, continuous delivery. Really, what we're 25 00:01:27,570 --> 00:01:29,640 going to spend about we're gonna spend a little time talking 26 00:01:29,640 --> 00:01:35,460 about is how you once your cluster is deployed, but not 27 00:01:35,460 --> 00:01:39,270 doing the actual development of the application. What happens in 28 00:01:39,270 --> 00:01:43,230 the middle, what tools are there out there to configure your 29 00:01:43,230 --> 00:01:46,380 cluster to set up all those security setup all the policies 30 00:01:46,380 --> 00:01:50,370 to set up all the configurations and the bits and the bytes, and 31 00:01:50,370 --> 00:01:53,100 get your cluster ready for applications and then deploy and 32 00:01:53,100 --> 00:01:54,870 manage those application life cycles? 33 00:01:55,680 --> 00:01:57,060 Tariq Islam: So it's interesting, you mentioned it 34 00:01:57,060 --> 00:02:03,210 that way, Jamie? Because it just kind of occurred to me that this 35 00:02:03,210 --> 00:02:09,270 is actually how we see a lot of organizations adopt things 36 00:02:09,300 --> 00:02:12,600 across the timeline. They Oh, it always starts with Okay, I'm 37 00:02:12,600 --> 00:02:15,690 gonna get my cluster stood up. But then how do I actually like 38 00:02:15,690 --> 00:02:20,310 manage this thing, day to ops is maybe synonymous with this 39 00:02:20,310 --> 00:02:24,360 topic, a little bit, right? cluster configuration, 40 00:02:24,360 --> 00:02:28,080 environment configuration, what does all that look like? And 41 00:02:28,080 --> 00:02:31,440 then they start worrying about creating a harness for for 42 00:02:31,440 --> 00:02:32,610 application delivery. 43 00:02:32,700 --> 00:02:34,710 John Osborne: And and maybe that that'll be the segue for us and 44 00:02:34,710 --> 00:02:38,730 to see ICD? It's been probably, to me, like the most fragmented 45 00:02:38,730 --> 00:02:42,570 space, right? I mean, we're like six years into Kubernetes. And 46 00:02:42,600 --> 00:02:45,780 there's been a ton of these projects that have just kind of 47 00:02:46,620 --> 00:02:49,410 been hot for like a year, and then he never really heard it 48 00:02:49,410 --> 00:02:50,220 from them again, right? 49 00:02:50,550 --> 00:02:52,470 Jamie Duncan: I mean, he was what, two and a half years ago, 50 00:02:52,470 --> 00:02:56,970 we we went over to tartes office in Northern Virginia, and looked 51 00:02:56,970 --> 00:02:58,710 at what case on it and 52 00:02:58,740 --> 00:02:59,670 John Osborne: chase on it. 53 00:03:00,269 --> 00:03:02,459 Jamie Duncan: json at ham sandwich sauna. I mean, there 54 00:03:02,459 --> 00:03:04,679 were a ton of them that we looked at, and I haven't heard 55 00:03:04,679 --> 00:03:08,789 of most of them. Since that day, when we stole free lunch from 56 00:03:08,789 --> 00:03:09,119 Google. 57 00:03:09,750 --> 00:03:11,940 Tariq Islam: I actually forgot it existed. 58 00:03:12,420 --> 00:03:13,050 Unknown: Exactly. 59 00:03:15,000 --> 00:03:17,550 Tariq Islam: Wow, man, it's been a while. I mean, this ecosystem 60 00:03:17,550 --> 00:03:20,340 is full like it's kind of defined as in the beginning 61 00:03:20,340 --> 00:03:24,990 there was Helm in Helm was pretty okay. Well, it fill the 62 00:03:24,990 --> 00:03:30,780 gap. It I mean, it it filled it filled a, an implicit need for 63 00:03:30,780 --> 00:03:33,840 any sis ops team, any sis ops team worth, it's worth its 64 00:03:33,840 --> 00:03:37,920 weight, right? If I'm gonna adopt Kubernetes, I really I 65 00:03:37,920 --> 00:03:42,420 need to be able to define define the gates, I need to define the 66 00:03:42,420 --> 00:03:48,990 guardrails and Helm did does a great job at that, of that, at 67 00:03:48,990 --> 00:03:52,770 least at some scale, but perhaps not at scale. And maybe we can 68 00:03:52,770 --> 00:03:57,030 talk about that today. But it definitely filled a need. I 69 00:03:57,030 --> 00:04:01,230 forgot how wild Kubernetes was in the early days until I went 70 00:04:01,230 --> 00:04:04,350 back to this episode and started looking at Helm to and like all 71 00:04:04,350 --> 00:04:07,710 this stuff came back like what Helm was first released, there 72 00:04:07,710 --> 00:04:11,760 was no our back in Kubernetes at all, and like and then how 73 00:04:11,760 --> 00:04:15,600 maintains like, this whole state, in its its own outside of 74 00:04:15,600 --> 00:04:18,090 Kubernetes, because there was no CR DS or anything like that. 75 00:04:18,090 --> 00:04:22,140 And, and everything used to run his route. And, you know, all 76 00:04:22,140 --> 00:04:25,770 the Docker images would mount to the home and the battles that 77 00:04:25,770 --> 00:04:30,840 were fought. Yeah, you know, so. So let's do this. Let's, let's, 78 00:04:30,840 --> 00:04:34,380 uh, let's define Helm real quick. For the folks listening, 79 00:04:34,410 --> 00:04:39,180 just, you know, for good podcast hygiene here. I'll take, I'll 80 00:04:39,180 --> 00:04:41,820 take a crack at it, it's probably gonna be half half 81 00:04:41,820 --> 00:04:43,410 wrong. What's gonna be half right? 82 00:04:44,430 --> 00:04:48,390 It'll be half right. I've always seen Helm as as, as a 83 00:04:48,390 --> 00:04:53,370 templating. approach to to. I almost have to say Git ops. 84 00:04:53,370 --> 00:04:54,960 Again, I don't want to overload that term. 85 00:04:55,020 --> 00:04:56,400 John Osborne: package management, maybe? 86 00:04:56,430 --> 00:04:59,400 Tariq Islam: Yes, exactly. It's application package, not 87 00:04:59,400 --> 00:05:01,020 application pack. Because remember treating your 88 00:05:01,020 --> 00:05:04,710 applications in a packaged way. But it's, it's it's a, it's a 89 00:05:04,710 --> 00:05:09,120 templating. It's a templating tool effectively. I mean, it's 90 00:05:09,120 --> 00:05:14,130 almost like a DSL. And it's very specific. And it is important to 91 00:05:14,130 --> 00:05:19,290 note by the way that that Helm is. It's It's different from 92 00:05:19,290 --> 00:05:22,200 customize. I'm introducing that term early on, because as soon 93 00:05:22,200 --> 00:05:27,570 as you Google Helm or customize, in your, in the Google quick 94 00:05:27,570 --> 00:05:30,750 results, you actually see, like one of the first results is Helm 95 00:05:30,750 --> 00:05:34,770 versus customize. So we'll talk about customize as well. But 96 00:05:34,770 --> 00:05:38,610 there seems to be a bit of a not a battle going on. But folks are 97 00:05:38,610 --> 00:05:42,750 wondering which one to use when and why hounding the first one 98 00:05:42,900 --> 00:05:47,310 to Market. But templating is how I would have I would describe 99 00:05:47,310 --> 00:05:47,550 it, 100 00:05:47,580 --> 00:05:50,220 John Osborne: that's how I think of it as I mean, to me, like my 101 00:05:50,220 --> 00:05:54,180 mental map, our model is I still kind of think of it like Jinja 102 00:05:54,180 --> 00:05:57,660 templates in Python, but that's probably shortchanging some of 103 00:05:57,780 --> 00:06:00,600 some of this. Yep. Similar capabilities. But 104 00:06:01,440 --> 00:06:04,170 Jamie Duncan: it really it feels a lot like To me it feels like 105 00:06:04,170 --> 00:06:05,310 Ansible for Kubernetes. 106 00:06:08,100 --> 00:06:12,690 Tariq Islam: Is it though, I'm channeling my, my Thor right 107 00:06:12,690 --> 00:06:12,900 now. 108 00:06:17,640 --> 00:06:21,240 Jamie Duncan: That's what. And as I was saying that I five 109 00:06:21,240 --> 00:06:23,850 seconds previously, because I, you know, in my brain out my 110 00:06:23,850 --> 00:06:27,750 mouth, I wonder if I feel that way, because that's when I 111 00:06:27,750 --> 00:06:30,720 really learned both of those tools. I learned them sort of 112 00:06:30,720 --> 00:06:34,950 beside one another. Or if there's a relationships there, I 113 00:06:34,950 --> 00:06:36,810 think there's at least some relationships there. Because 114 00:06:36,810 --> 00:06:39,600 john, you know, explicitly called out Jinja templates, 115 00:06:39,600 --> 00:06:41,490 which is how you do templates in Ansible. 116 00:06:41,880 --> 00:06:42,330 Unknown: Mm hmm. 117 00:06:42,510 --> 00:06:44,070 Jamie Duncan: there on the surface, there are at least some 118 00:06:44,070 --> 00:06:48,630 similarities. But what Ansible is, at the end of the day is 119 00:06:48,630 --> 00:06:52,170 largely a templating engine. So they have 120 00:06:52,200 --> 00:06:54,180 Tariq Islam: maybe a right that actually, if you're going by 121 00:06:54,180 --> 00:06:57,810 that description, because that's how I mean, that's how that's 122 00:06:57,810 --> 00:07:04,080 how we see helm. Now. I mean, that being said, though, I think 123 00:07:04,080 --> 00:07:06,690 there's there's a discussion happening right now around 124 00:07:06,690 --> 00:07:11,910 whether whether a, an ops team that's managing this Kubernetes 125 00:07:11,910 --> 00:07:16,890 platform should use Helm or customize is the room to use 126 00:07:16,890 --> 00:07:21,600 both. Maybe I'll nip that last one in the bud right now and say 127 00:07:21,600 --> 00:07:24,870 using both is just going to add way too much complexity. But I 128 00:07:24,870 --> 00:07:27,420 think it's a good discussion. Because if you if you look at 129 00:07:27,420 --> 00:07:30,810 the the structure of helm, I mean, it's come a long way in 130 00:07:30,810 --> 00:07:32,910 terms of its architecture as well. They look at security 131 00:07:32,910 --> 00:07:37,020 concerns, initially, you had client side, server side 132 00:07:37,050 --> 00:07:40,800 components, and now you don't, they've done a great job of 133 00:07:40,800 --> 00:07:43,560 locking certain things down and addressing the security concerns 134 00:07:43,560 --> 00:07:44,160 that were 135 00:07:44,310 --> 00:07:46,770 John Osborne: that were there early on. If you were in the in 136 00:07:46,770 --> 00:07:49,410 the early days, you know, they have these in the old version of 137 00:07:49,410 --> 00:07:52,230 Helm a this thing called tiller, which basically, yeah, it's 138 00:07:52,230 --> 00:07:54,330 mashed up all the template was basically was the templating 139 00:07:54,330 --> 00:07:57,990 engine and would just apply everything to the Kubernetes 140 00:07:57,990 --> 00:08:00,180 API, there was no security because it didn't exist in 141 00:08:00,180 --> 00:08:02,850 Kubernetes. But do you actually remember the beginning? And I 142 00:08:02,850 --> 00:08:05,220 forgot about this until I started going back and look at 143 00:08:05,220 --> 00:08:08,820 some wild stuff, but it would actually just deploy everything 144 00:08:08,820 --> 00:08:10,950 in the coop system namespace. 145 00:08:12,300 --> 00:08:13,290 Jamie Duncan: That's all it would do. 146 00:08:14,160 --> 00:08:15,810 John Osborne: Well, that was like the default namespace it 147 00:08:15,810 --> 00:08:19,590 would deploy to. So yeah, so he was not iterating. And all of a 148 00:08:19,590 --> 00:08:22,530 sudden, you're just not really, yeah. Which is a really good way 149 00:08:22,530 --> 00:08:29,550 to like FUBAR your cluster, but I do like I do like helm. It did 150 00:08:29,550 --> 00:08:32,250 fill a gap in the beginning. But if you also look at a lot of the 151 00:08:32,250 --> 00:08:35,820 tools that have been around, it's the only one, like, 152 00:08:35,820 --> 00:08:38,610 customize as much as someone new right, like out of the first 153 00:08:38,610 --> 00:08:41,550 generation of tools. It's the only way to hell yeah, still 154 00:08:41,550 --> 00:08:42,180 here. 155 00:08:43,379 --> 00:08:46,349 Tariq Islam: So john, and your and so if I were to ask you the 156 00:08:46,349 --> 00:08:49,049 question, I don't think we've addressed this for for the, for 157 00:08:49,049 --> 00:08:53,099 the folks listening, what what do I use Helm for? 158 00:08:55,140 --> 00:08:59,670 John Osborne: I always, so two parts that I always felt like, 159 00:09:00,330 --> 00:09:03,060 especially in the early days of Kubernetes, but still probably 160 00:09:03,060 --> 00:09:05,790 to this day, one of the big things about it was actually the 161 00:09:05,790 --> 00:09:09,030 content around it. So there's like a Helm chart that exists 162 00:09:09,030 --> 00:09:11,550 for almost anything. So if you want to go install MySQL on 163 00:09:11,550 --> 00:09:13,860 Kubernetes, or, or, 164 00:09:14,939 --> 00:09:19,679 you know, some Apache project or, or WordPress, because every 165 00:09:19,679 --> 00:09:22,709 example is WordPress, right? You can just go you can just type 166 00:09:22,709 --> 00:09:26,399 Helm install, and it'll pull in a chart that has that for you. 167 00:09:27,179 --> 00:09:30,299 Now, there's of course, like, you know, there's security 168 00:09:30,299 --> 00:09:32,909 concerns around that. There's, there's ways to do that more 169 00:09:32,909 --> 00:09:35,669 securely, but to get up and running the content around it, 170 00:09:35,699 --> 00:09:38,639 almost like Docker, right? Like if you think about when Docker 171 00:09:38,639 --> 00:09:41,579 was really big. In the beginning, it was more. There's 172 00:09:41,579 --> 00:09:44,009 all these based, yes, they did everything with API, all that 173 00:09:44,009 --> 00:09:46,349 stuff. But I think a big reason was they have the ecosystem, 174 00:09:46,349 --> 00:09:50,789 right, and Cal has the ecosystem. But I don't think it 175 00:09:50,789 --> 00:09:56,339 works that well for individual use cases. But it does work well 176 00:09:56,339 --> 00:09:59,849 for teams that are going to gate. At least that's where I 177 00:09:59,849 --> 00:10:01,979 start. See it. And that's where I see people using it as more 178 00:10:01,979 --> 00:10:05,429 like, you know, ops teams or some, some team that's going to 179 00:10:05,429 --> 00:10:07,649 gate resources to a developer, but they're going to give them 180 00:10:07,649 --> 00:10:11,159 access to certain variables, then you can, they can give them 181 00:10:11,159 --> 00:10:14,189 Helm charts that they can override certain things, but not 182 00:10:14,189 --> 00:10:19,019 certain other things. I also see kind of, sometimes in the 183 00:10:19,019 --> 00:10:23,129 enterprise teams will have like, they'll allow, say, like five 184 00:10:23,129 --> 00:10:25,859 variations on let's use like Tomcat as a simple example. But 185 00:10:26,729 --> 00:10:29,999 they're going to allow you five, five variations of Tomcat rate 186 00:10:29,999 --> 00:10:33,749 and Helm is a kind of a good way to lock down Tomcat, but just 187 00:10:33,749 --> 00:10:35,939 give you the five variations, and you can kind of pick and 188 00:10:35,939 --> 00:10:37,559 choose with the templating. 189 00:10:39,899 --> 00:10:41,519 Tariq Islam: Yeah, and so that's, that's where I feel like 190 00:10:41,519 --> 00:10:45,089 Helm starts to come apart a bit, because of all the possible 191 00:10:45,089 --> 00:10:49,109 variations that you start running into as an organization. 192 00:10:49,199 --> 00:10:52,169 You know, if I've got five variations, you know, in a 193 00:10:52,169 --> 00:10:54,059 couple years, I might end up becoming 15 different 194 00:10:54,059 --> 00:10:58,169 variations. And I think that's where the discussion with 195 00:10:58,199 --> 00:11:01,649 between Helm and customize comes into play. Do I want to do like 196 00:11:01,649 --> 00:11:04,589 templating till till I die? Or do I want to actually start 197 00:11:04,829 --> 00:11:08,369 leveraging an overlay based system? Like the one customize 198 00:11:08,369 --> 00:11:12,629 represents? I mean, we haven't we haven't even defined 199 00:11:12,629 --> 00:11:14,819 customize. Yep. But I think that's that's the question. 200 00:11:14,879 --> 00:11:16,349 John Osborne: Do you want to define customers? 201 00:11:16,590 --> 00:11:20,130 Tariq Islam: Yeah, so so customize. It's not it's not 202 00:11:20,130 --> 00:11:24,090 templating? It's primarily it's, it isn't work, for lack of a 203 00:11:24,090 --> 00:11:28,080 better term, an overlay engine, right. And the unique thing 204 00:11:28,080 --> 00:11:31,440 about customize is that it's actually baked into the cube 205 00:11:31,440 --> 00:11:35,790 CTL. See, ally. So you can use the dash k flag to use it. from 206 00:11:35,790 --> 00:11:38,790 an engineering standpoint, it's it's maintained by, you know, 207 00:11:38,790 --> 00:11:43,410 core Kubernetes, whereas Helm is, is maintained by by a third 208 00:11:43,410 --> 00:11:45,600 party, said, distributors, what 209 00:11:45,600 --> 00:11:48,330 John Osborne: is the third? I mean, it seems mostly IBM and 210 00:11:48,420 --> 00:11:50,190 Microsoft, from what I've seen, but 211 00:11:50,970 --> 00:11:52,950 Tariq Islam: yeah, that's what I've seen as well. That's what I 212 00:11:52,950 --> 00:11:55,260 mean by third party. I'm not saying that it's outside the 213 00:11:55,260 --> 00:11:59,520 community. But the fact that customize is part of you know, 214 00:11:59,520 --> 00:12:06,060 the coop CTL. See, Li like the that actual flow, just kind of 215 00:12:06,060 --> 00:12:09,300 intrinsically brings it closer to the Kubernetes platform. And 216 00:12:09,300 --> 00:12:12,990 I'm, I've always been a bit of a purist slash, minimalist slash, 217 00:12:12,990 --> 00:12:15,870 whatever, just, I like that type of thing, right? Just bring me 218 00:12:15,870 --> 00:12:18,150 up, keep me as close to Kubernetes as you as you 219 00:12:18,150 --> 00:12:19,470 possibly can. And I'm happy. 220 00:12:20,070 --> 00:12:22,650 John Osborne: Tell me if my mental map is just horrible, but 221 00:12:22,680 --> 00:12:26,850 in my mind, I kind of think of customizes. Like, if I was 222 00:12:26,850 --> 00:12:30,000 working locally, and sort of scripting things out with sed 223 00:12:30,000 --> 00:12:33,660 and a bunch of other Linux command line tools, eventually, 224 00:12:34,230 --> 00:12:36,720 when I that mature it, I get something like customize. 225 00:12:37,259 --> 00:12:38,159 Tariq Islam: That's about right, man. 226 00:12:38,429 --> 00:12:41,309 Jamie Duncan: Yeah, that's, yeah, you're certainly in the 227 00:12:41,309 --> 00:12:45,059 right neighborhood, if not at the right front door. It can 228 00:12:45,059 --> 00:12:45,389 turn on 229 00:12:45,450 --> 00:12:46,110 John Osborne: their eyes. 230 00:12:49,110 --> 00:12:53,640 Jamie Duncan: Right, like Tarik said there, you can overlay. So 231 00:12:53,670 --> 00:12:59,670 you can you have a directory called base. And inside base are 232 00:12:59,670 --> 00:13:02,730 all of the objects that you want to have created when you run 233 00:13:02,730 --> 00:13:04,890 this customize script. So if you want to create a namespace, 234 00:13:05,670 --> 00:13:08,640 deployment objects, Ingress objects, whatever, and you can 235 00:13:08,640 --> 00:13:12,990 parameterize those. So in that regard, it's not drastically 236 00:13:12,990 --> 00:13:17,730 different than helm. And they even look a little alike. So you 237 00:13:17,730 --> 00:13:22,230 have this base directory, and then everything you run, pull 238 00:13:22,230 --> 00:13:25,680 starts at that base directory. But then you can overlay 239 00:13:25,680 --> 00:13:28,080 different things on top of that. So you can also have a directory 240 00:13:28,080 --> 00:13:31,470 called production or a directory called development. And you can 241 00:13:31,470 --> 00:13:35,490 populate those parameters with directory with dev values, or 242 00:13:35,490 --> 00:13:38,910 prod values or your own custom values. And then you can also 243 00:13:38,910 --> 00:13:41,790 have parameters files, where you're just filling things in, 244 00:13:41,940 --> 00:13:46,080 or where you override a specific thing. It gets to the same 245 00:13:46,080 --> 00:13:50,460 place, my biggest, the reason I use customize more than I use 246 00:13:50,460 --> 00:13:57,000 Helm is that I keep everything in the repository. The biggest 247 00:13:57,150 --> 00:14:02,250 problem I have with Helm is that I have to pull in, I have to do 248 00:14:02,250 --> 00:14:07,350 Helm repo add, and then go add the repository for WordPress 249 00:14:07,860 --> 00:14:10,980 that contains the helm charts, or for whatever I'm installing. 250 00:14:11,460 --> 00:14:14,160 And then I have to turn around and tell it to go update those 251 00:14:14,160 --> 00:14:19,770 like, like a package manager, like using Yum, or apt. And then 252 00:14:19,770 --> 00:14:22,320 they're stored off in some library folder. And they're not 253 00:14:22,350 --> 00:14:27,000 a part of my repository, where is customize inside the 254 00:14:27,000 --> 00:14:29,910 Customize library itself. It just understands what a 255 00:14:29,910 --> 00:14:33,810 namespace, what that yamo looks like. It understands all these 256 00:14:33,810 --> 00:14:36,750 different things. And I can do other cool things like for 257 00:14:36,750 --> 00:14:39,600 secrets, I can actually just have it I can feed it two or 258 00:14:39,600 --> 00:14:42,270 three pieces of information and have it generate a secret for 259 00:14:42,270 --> 00:14:45,180 me. You do a lot of really cool little shortcut, things like 260 00:14:45,180 --> 00:14:50,280 that inside customize. And it's all inside my repo. I don't have 261 00:14:50,280 --> 00:14:53,550 to pull another another chunk of code down and put like, like 262 00:14:53,550 --> 00:14:55,020 subscribing to a repository. 263 00:14:55,740 --> 00:14:57,840 Tariq Islam: And I think that's part of the power right for an 264 00:14:57,840 --> 00:15:02,760 organization to be able to start adopting a thing like, like get 265 00:15:02,760 --> 00:15:04,890 ups, right? I mean, if you're keeping everything in 266 00:15:04,890 --> 00:15:08,220 repository, it kind of lends itself towards that. So 267 00:15:08,220 --> 00:15:10,890 immediately, I think it becomes a lot more scalable. Now, for 268 00:15:10,890 --> 00:15:15,060 me, I still am a fan of helm. Because it I mean, if you've 269 00:15:15,060 --> 00:15:17,550 installed this do in the last year, you probably use the helm 270 00:15:17,550 --> 00:15:21,090 install or the helm chart for it. I love that right? I love 271 00:15:21,090 --> 00:15:22,800 being able to just go to Helm chart and install something, 272 00:15:22,800 --> 00:15:25,440 it's pre configured pre package ready to rock. That's been 273 00:15:25,440 --> 00:15:29,250 great. But But to your point, Jamie, it's I lean more towards 274 00:15:29,280 --> 00:15:32,880 towards customize for for things that I want to break apart and 275 00:15:32,880 --> 00:15:34,710 put back together for my own purposes, 276 00:15:34,830 --> 00:15:41,340 John Osborne: it does seem like in terms of ci CD usage, that I 277 00:15:41,340 --> 00:15:45,240 see more customized now, in terms of, you know, someone's 278 00:15:45,240 --> 00:15:48,060 going to write some code or it's going to iterate I know in the 279 00:15:48,060 --> 00:15:51,270 Git ops model, like the default projects like Argo and flux, 280 00:15:51,270 --> 00:15:56,280 like most of their examples, use customized as well. I still see 281 00:15:56,280 --> 00:16:00,330 how I'm filling a need. In terms of the the teams and the people 282 00:16:00,330 --> 00:16:03,690 in that handoff, do you see customize being used for those 283 00:16:03,690 --> 00:16:04,980 cases as well? or? 284 00:16:05,879 --> 00:16:06,449 Unknown: Yeah, 285 00:16:06,480 --> 00:16:11,610 Jamie Duncan: Yeah, I do. And because of that overlay idea 286 00:16:11,610 --> 00:16:15,990 that I can, I have a folder that contains all of the uniqueness 287 00:16:15,990 --> 00:16:19,020 that represents my production environment. And it's just 288 00:16:19,020 --> 00:16:22,230 labeled production, or a dev a dev environment or, and I can 289 00:16:22,230 --> 00:16:25,800 split it and slice it and dice it however I want. And it's 290 00:16:25,800 --> 00:16:30,090 going to take those objects in, in my base folder, my base 291 00:16:30,090 --> 00:16:34,200 directory, and it's going to add or subtract from them, as per 292 00:16:34,200 --> 00:16:38,700 the instructions in those in the customized folders. So I can, I 293 00:16:38,700 --> 00:16:43,680 can get to that same place. And I can do all of the things that 294 00:16:43,680 --> 00:16:47,520 you're supposed to do inside there. And it's all and kind of 295 00:16:47,520 --> 00:16:50,940 interesting, when you're talking about where it's the default for 296 00:16:50,940 --> 00:16:56,220 ci CD. And I think it's just because it's easier is because 297 00:16:56,220 --> 00:17:00,330 it's part of coop CTL when I install coop CTL I install 298 00:17:00,330 --> 00:17:04,740 customize all of your container images that do all of those 299 00:17:04,740 --> 00:17:09,210 builds have coop CTL in them because they have to, as so and 300 00:17:09,240 --> 00:17:12,360 so I just get all of that functionality out of the box. 301 00:17:12,390 --> 00:17:15,450 Whereas if I'm using helm, every time I run that build my 302 00:17:15,450 --> 00:17:18,630 container image has to grab the repo and pull the repo down and 303 00:17:18,630 --> 00:17:23,100 do all of the things you have to do to make him work. And that's 304 00:17:23,130 --> 00:17:26,730 time consuming at scale when I'm doing lots of build today. 305 00:17:27,329 --> 00:17:29,399 John Osborne: So how am I think one of the reasons why Helm is 306 00:17:29,399 --> 00:17:32,819 so popular just has this massive ecosystem of existing charts, 307 00:17:32,819 --> 00:17:35,519 right? I mean, is there ways to plug those charts in to 308 00:17:35,519 --> 00:17:39,659 customize or kind of leverage that ecosystem? Or or do you 309 00:17:39,689 --> 00:17:42,209 have to use Helm if you want to leverage all that? 310 00:17:43,050 --> 00:17:45,840 Tariq Islam: You'd have to you'd have to pull apart the the helm 311 00:17:45,840 --> 00:17:51,240 chart a little bit. Because it's in its its its own non k 312 00:17:51,240 --> 00:17:54,900 standard format. It's still YAML. But it's still its own. 313 00:17:54,930 --> 00:17:57,810 That's what I call it. That's why I mentioned it kind of being 314 00:17:57,840 --> 00:18:01,800 its own DSL earlier. Because you really it if you're on Helm 315 00:18:01,800 --> 00:18:03,840 today, you've been on Helm for years and you want to go over to 316 00:18:03,840 --> 00:18:07,710 customize, it's going to be a migration. That's what you're 317 00:18:07,740 --> 00:18:10,950 effectively looking at. Right? It's some form of a migration. 318 00:18:12,030 --> 00:18:14,820 John Osborne: I do like how I customize, it doesn't change the 319 00:18:14,820 --> 00:18:19,110 original YAML that you had. Makes it Yeah, just makes it 320 00:18:19,110 --> 00:18:22,110 easier. What about in production? Right? Like, is it? 321 00:18:23,010 --> 00:18:26,040 Do you think there's benefits to one versus the other? if, you 322 00:18:26,040 --> 00:18:29,970 know, some say some YAML goes astray, right? Like, what's the 323 00:18:30,150 --> 00:18:33,000 Is there a way you can tell inside the annotations or inside 324 00:18:33,000 --> 00:18:37,680 the metadata? How that yamo was constructed in the first place? 325 00:18:38,430 --> 00:18:40,950 Tariq Islam: I don't think that there's any auditing, how do you 326 00:18:40,950 --> 00:18:41,970 unpack that? I know and 327 00:18:42,690 --> 00:18:45,090 Jamie Duncan: I know, customize has like a dry run option. In 328 00:18:45,090 --> 00:18:50,070 fact, the default, yeah, just output YAML. So the way you 329 00:18:50,070 --> 00:18:52,590 actually use customize, you know, the customized command 330 00:18:53,160 --> 00:18:57,210 reads a dot customize file that then you tell it you feed it 331 00:18:57,210 --> 00:19:00,510 those parameters, but by default, it just spits out YAML 332 00:19:01,290 --> 00:19:04,860 and then you actually pipe that into if you're not using coop 333 00:19:04,860 --> 00:19:09,690 CTL dash k. a bad habit that I formed early on and still do is 334 00:19:09,690 --> 00:19:13,230 to run the customized command and then pipe that into coop 335 00:19:13,230 --> 00:19:13,830 CTL. 336 00:19:14,460 --> 00:19:16,290 Tariq Islam: You don't even do the dry run, you just fly by the 337 00:19:16,290 --> 00:19:16,470 seat. 338 00:19:17,280 --> 00:19:20,250 Jamie Duncan: Exactly like you get pipe 339 00:19:22,140 --> 00:19:23,550 John Osborne: live production, 340 00:19:24,780 --> 00:19:26,610 Unknown: exact reference or for the week. 341 00:19:28,290 --> 00:19:30,000 Jamie Duncan: Dry runs for people with SLA. 342 00:19:33,150 --> 00:19:35,430 John Osborne: Well, yeah, and Kubernetes does have some of 343 00:19:35,430 --> 00:19:39,900 that built in with the the last apply configuration annotation 344 00:19:39,900 --> 00:19:42,120 that's in there. Now if you look in I think it's 118. They added 345 00:19:42,120 --> 00:19:47,160 that where you can go see what the previous yamo was before it 346 00:19:47,160 --> 00:19:48,720 got applied. 347 00:19:49,440 --> 00:19:51,300 Jamie Duncan: I don't know there's any way The first time I 348 00:19:51,300 --> 00:19:51,840 saw it. 349 00:19:52,650 --> 00:19:55,080 John Osborne: I don't care for I don't care for it. I saw in the 350 00:19:55,080 --> 00:19:58,380 Git repo that they people complained and they said that 351 00:19:58,380 --> 00:20:00,450 people like it because it provides info but I just feel 352 00:20:00,450 --> 00:20:03,630 like it takes so much space. That to diverge too much, but 353 00:20:05,160 --> 00:20:07,320 Tariq Islam: religious c li best practices. 354 00:20:07,349 --> 00:20:12,089 John Osborne: Yeah. But I didn't know if there's any way to I'm 355 00:20:12,089 --> 00:20:15,959 not it. I'm not expert enough in either of them even though I've 356 00:20:15,959 --> 00:20:20,729 used both them a decent amount, but to actually unpack the how 357 00:20:20,819 --> 00:20:24,299 they were created. So you could see where things went awry. 358 00:20:24,750 --> 00:20:27,450 Jamie Duncan: Yeah, I don't think there is like it unless 359 00:20:27,450 --> 00:20:31,560 you tell it to. So I have significantly like, I'm very 360 00:20:31,560 --> 00:20:36,090 much a pure consumer of helm, where I do exactly what you 361 00:20:36,090 --> 00:20:39,000 said, john, I, you know, I'm reading the instructions on how 362 00:20:39,000 --> 00:20:41,160 to deploy something that I haven't deployed before. And it 363 00:20:41,160 --> 00:20:44,220 says, Go grab the helm chart and run Helm and set these set these 364 00:20:44,220 --> 00:20:48,420 variables. I do exactly. That's, that's pretty much my experience 365 00:20:48,420 --> 00:20:52,770 with helm. Customize, I've gone a fair bit deeper, I'm certainly 366 00:20:52,770 --> 00:20:55,590 not an expert by any stretch, but have, you know, sort of 367 00:20:55,590 --> 00:20:58,680 written some customized schemas and done some work to deploy to 368 00:20:58,680 --> 00:21:03,360 different environments. So I can say pretty confidently, like, in 369 00:21:03,360 --> 00:21:06,420 customized, there's not an inbuilt function to let you do 370 00:21:06,420 --> 00:21:11,250 that. But if, but that idea of everything's a parameter, 371 00:21:11,250 --> 00:21:15,210 anything I want is a parameter. I could add, you know, 372 00:21:15,270 --> 00:21:17,760 annotations and labels and all the things that I would want to 373 00:21:17,760 --> 00:21:23,310 add. to to see any of that information. I don't know that I 374 00:21:23,310 --> 00:21:26,970 like the idea of debugging that way. But that's kind of Yeah, I 375 00:21:26,970 --> 00:21:28,290 know what you're getting to maybe, 376 00:21:28,470 --> 00:21:31,680 John Osborne: yeah, just you know, any other information, you 377 00:21:31,680 --> 00:21:35,160 know, it's, uh, I'm always, I'm always looking for how I can 378 00:21:35,160 --> 00:21:38,070 unpack things, you know, how things gotten to the state, they 379 00:21:38,070 --> 00:21:40,140 got into any any information, there's good. 380 00:21:40,649 --> 00:21:43,529 Tariq Islam: So Helm was great initially, because it filled 381 00:21:43,529 --> 00:21:47,069 that gap, right, it filled that need for, you know, establishing 382 00:21:47,069 --> 00:21:52,259 those guardrails, but I also felt like it violated, you know, 383 00:21:52,259 --> 00:21:58,769 certain principles around DevOps and actually brought, it brought 384 00:21:58,769 --> 00:22:01,379 certain things too far over to the right, I guess. So, you 385 00:22:01,379 --> 00:22:05,459 know, I, I interacted with organizations that had adopted 386 00:22:05,459 --> 00:22:09,719 Helm and suddenly had developers using Helm charts and kind of 387 00:22:09,719 --> 00:22:13,529 overreaching into the infrastructure deployment 388 00:22:13,529 --> 00:22:17,909 process, application deployment process, and it muddied the 389 00:22:17,909 --> 00:22:22,229 waters quite a bit. And it, it actually prevented operations 390 00:22:22,229 --> 00:22:27,779 teams from being able to make certain decisions without, you 391 00:22:27,779 --> 00:22:31,709 know, significantly impacting everybody else around them, 392 00:22:31,859 --> 00:22:34,529 which is kind of the antithesis of why they were trying to do 393 00:22:34,529 --> 00:22:38,939 all of this to begin with. And I found that that customized 394 00:22:38,939 --> 00:22:43,049 somewhat addresses that because it's not customized is just not 395 00:22:43,049 --> 00:22:49,589 something that a developer really can worry about, I guess, 396 00:22:49,589 --> 00:22:54,119 given where customize lives, in the process of creating these 397 00:22:54,119 --> 00:22:58,469 overlays and applying them at a given moment in time, 398 00:22:59,250 --> 00:23:02,100 Jamie Duncan: right, they could do it in on their laptop, they 399 00:23:02,100 --> 00:23:04,950 could you know, the, the customized files are in the 400 00:23:04,950 --> 00:23:08,670 repository. But they couldn't do it in an environment that 401 00:23:08,670 --> 00:23:13,140 mattered because our back because of all that stuff that 402 00:23:13,140 --> 00:23:18,240 didn't exist back in the day, to take care of it. Customize is 403 00:23:18,240 --> 00:23:22,110 able to take advantage of it, I think, as I look at and we kind 404 00:23:22,110 --> 00:23:27,030 of dogged, not dogged him, but we certainly, at least I gave it 405 00:23:27,480 --> 00:23:31,230 a poor connotation when we were talking about all the service 406 00:23:31,230 --> 00:23:35,040 meshes and how hashey corpse console deploys with the helm 407 00:23:35,040 --> 00:23:38,610 chart. And I said it felt kind of like 2005 408 00:23:39,390 --> 00:23:40,320 Tariq Islam: I remember that. 409 00:23:42,000 --> 00:23:46,020 Jamie Duncan: Like that is like that's with a slightly more 410 00:23:46,020 --> 00:23:50,310 intelligent thought on it. I hope customize is taking the 411 00:23:50,310 --> 00:23:56,640 lessons learned from helm. And building. prob, for the reasons 412 00:23:56,640 --> 00:23:59,970 you just said tarak, as well as the the stuff I was talking 413 00:23:59,970 --> 00:24:05,820 about, like where it's, it's easier for it to live in a 414 00:24:05,820 --> 00:24:09,930 repository without any external dependencies. It's taking the 415 00:24:09,930 --> 00:24:13,140 lessons learned from Helm and making it a more cloud native 416 00:24:13,140 --> 00:24:16,020 DevOps friendly get ops friendly, buzzword friendly 417 00:24:16,170 --> 00:24:16,680 tool. 418 00:24:17,130 --> 00:24:20,760 Tariq Islam: Yeah, I just think and this is not to say that Helm 419 00:24:20,760 --> 00:24:24,690 had poor design principles that they certainly don't, they never 420 00:24:24,690 --> 00:24:28,380 did, but I think they fell victim to having to do things 421 00:24:28,380 --> 00:24:32,970 that I think at this point, you know, five years later, we know 422 00:24:32,970 --> 00:24:36,780 that we probably want to leave to some other tool or to some 423 00:24:36,780 --> 00:24:40,620 other thing and doing it some other way. And I think 424 00:24:40,620 --> 00:24:44,610 customises has benefited from the things that Helm had to just 425 00:24:44,610 --> 00:24:48,420 deal with. Right, so Helm is kind of like, you know, back in 426 00:24:48,420 --> 00:24:52,650 my day, right, we had to walk to school and shoes and six foot 427 00:24:52,650 --> 00:24:56,670 snow and work without shoes. It's that type of thing. I just 428 00:24:56,670 --> 00:24:59,520 feel like customized has taken has been able to take advantage 429 00:24:59,520 --> 00:25:05,850 of of just being that next iteration, I guess. I mean, 430 00:25:05,850 --> 00:25:10,260 maybe that's the conclusion that we're kind of alluding to here 431 00:25:10,920 --> 00:25:13,590 between the two. But that's not also not to say that you 432 00:25:13,590 --> 00:25:18,210 shouldn't use helm. Can we? Can I Can any of us say that, that 433 00:25:18,210 --> 00:25:21,840 there isn't a use case today? Where I 434 00:25:21,839 --> 00:25:23,789 John Osborne: think I think that's where I'm struggling to, 435 00:25:23,789 --> 00:25:27,569 because Helm still has a Mac, ball. And also kudos to the home 436 00:25:27,569 --> 00:25:31,409 team for what they did with with three, they did. A lot of that 437 00:25:31,409 --> 00:25:34,649 complexity, they ripped out Teller, they moved it all into 438 00:25:34,649 --> 00:25:38,669 the, you know, into the client side, their mood, much code, all 439 00:25:38,669 --> 00:25:42,719 that stuff. So kudos to them. But they, they do still have a 440 00:25:42,719 --> 00:25:48,239 massive ecosystem of users. And even though I use Helm when it 441 00:25:48,239 --> 00:25:52,289 first came out, with helm, to which I think one was out for 442 00:25:52,289 --> 00:25:55,949 like a month or something at the time, but it does kind of feel 443 00:25:55,949 --> 00:26:02,189 like that most of the users are not net you new users. And maybe 444 00:26:02,189 --> 00:26:06,479 that's wrong, but their ecosystem is so big. So maybe 445 00:26:06,479 --> 00:26:10,259 I'd like to wish we had somebody on the show that that just 446 00:26:10,259 --> 00:26:12,809 picked a new that just picked a new tool, and they picked him 447 00:26:12,809 --> 00:26:15,929 over customized. And maybe it is that that people and that the 448 00:26:15,929 --> 00:26:18,449 templating. And maybe that mental model just works better 449 00:26:18,449 --> 00:26:19,859 for them than having 450 00:26:21,240 --> 00:26:23,310 Jamie Duncan: an interesting thought I just had as we were as 451 00:26:23,310 --> 00:26:29,460 y'all were talking that through. If I'm building an application, 452 00:26:29,700 --> 00:26:33,300 and I'm the ops team, and I have a dev team or multiple dev 453 00:26:33,300 --> 00:26:39,420 teams, and I'm deploying it into my infrastructure, I think 454 00:26:39,420 --> 00:26:44,580 customized is easier, is a more effective solution. But if I'm 455 00:26:45,360 --> 00:26:49,980 sending something out to the public, as a product, I have 456 00:26:49,980 --> 00:26:54,210 Helm chart is what I would most likely pick. And I don't think 457 00:26:54,240 --> 00:26:59,250 it's just because of the size of the ecosystem. I think those 458 00:26:59,250 --> 00:27:02,490 Helm charts with the external dependencies that I can control 459 00:27:02,550 --> 00:27:08,640 as a product owner, make that workflow a lot easier than 460 00:27:08,640 --> 00:27:12,540 saying go grab some GitHub, and here's the customized file. And 461 00:27:12,540 --> 00:27:17,310 here's the variables you can override it let's it is it is 462 00:27:17,310 --> 00:27:20,520 packaging my my container native application. 463 00:27:21,060 --> 00:27:23,910 Tariq Islam: Yeah, no, maybe maybe that maybe maybe there's 464 00:27:23,910 --> 00:27:26,550 absolutely room for both, right, because I've always looked at 465 00:27:26,550 --> 00:27:32,910 customize. Something that's more infrastructure related, 466 00:27:33,090 --> 00:27:36,330 environmental related. So if I'm trying to define to your point, 467 00:27:36,330 --> 00:27:38,880 Jamie earlier, my dev environment, my production 468 00:27:38,880 --> 00:27:43,560 environment, everything in between. and, and leverage the 469 00:27:43,560 --> 00:27:46,710 overlay architecture that customized provides me to be 470 00:27:46,710 --> 00:27:50,280 able to do that at scale. That's That's me harnessing things 471 00:27:50,280 --> 00:27:52,980 right or setting up a harness. And those are my guardrails. And 472 00:27:52,980 --> 00:27:56,430 it's easy for me to do at scale, especially in enterprises, where 473 00:27:56,640 --> 00:27:58,890 there's like maybe dozens and dozens of lines of businesses 474 00:27:58,890 --> 00:28:01,860 that I've got to be able to support. How long the other hand 475 00:28:02,220 --> 00:28:07,230 would, I guess be further left, where we're to now talking 476 00:28:07,230 --> 00:28:12,570 about? Well, we, we support or approve this specific 477 00:28:12,570 --> 00:28:15,450 configuration of my sequel to be deployed? 478 00:28:16,859 --> 00:28:19,019 John Osborne: I think to Jamie kind of touched on something and 479 00:28:19,019 --> 00:28:23,159 I'm not even sure how to articulate it. But I feel like 480 00:28:23,849 --> 00:28:29,519 customize makes assumptions about your ops team. And I think 481 00:28:29,519 --> 00:28:33,599 that there's something there with with Git Ops, where Git ops 482 00:28:33,599 --> 00:28:36,779 makes assumptions that it's not just infrastructure as code, but 483 00:28:36,779 --> 00:28:39,599 your workflows revolve around Git. And I feel like with 484 00:28:39,599 --> 00:28:43,529 customize, it's making assumptions that your ops team 485 00:28:43,589 --> 00:28:48,299 has workflows that are good, that are focused around get, and 486 00:28:48,389 --> 00:28:51,449 that you're working really closely with your dev teams, 487 00:28:51,449 --> 00:28:54,539 which maybe isn't always the case. And I don't know if I've 488 00:28:54,539 --> 00:28:56,579 articulated that correctly. But I feel like there's there's 489 00:28:56,579 --> 00:28:59,159 something there with the assumptions around customize and 490 00:28:59,159 --> 00:29:01,739 how you're working. It's the design principles. You're 491 00:29:01,740 --> 00:29:04,680 Tariq Islam: right. And I think, and I just made another 492 00:29:04,770 --> 00:29:09,450 Association, as you were saying that I I've always felt like and 493 00:29:09,450 --> 00:29:11,670 I don't know, I'm just not realizing this, but I've always 494 00:29:11,670 --> 00:29:16,980 felt like Helm charts. Were very similar to openshift templates. 495 00:29:18,240 --> 00:29:19,500 Is that a fair analogy? 496 00:29:20,370 --> 00:29:21,150 Unknown: Yeah. 497 00:29:21,180 --> 00:29:21,960 Tariq Islam: Do you guys remember that? 498 00:29:21,990 --> 00:29:27,810 John Osborne: Yeah. So openshift templates did have some extra 499 00:29:28,020 --> 00:29:31,380 things in there. But yeah, it was probably 80% analogous. 500 00:29:32,880 --> 00:29:36,210 Tariq Islam: Just in terms of use cases? Yeah. Not so like for 501 00:29:36,210 --> 00:29:40,800 like, but just giving, giving, you know, they didn't on the 502 00:29:40,800 --> 00:29:46,860 right. And a way to give some sort of well defined, not 503 00:29:46,890 --> 00:29:49,740 necessarily catalog of thing that's something separate No, I 504 00:29:49,740 --> 00:29:53,400 think that you know, if you want to if you want to deploy a 505 00:29:53,400 --> 00:29:57,210 patchy if you want to deploy my sequel if you want to deploy sto 506 00:29:57,210 --> 00:30:01,230 if you want to deploy X, Y or Z thing on top of this hardness or 507 00:30:01,230 --> 00:30:04,620 on top of this environment that I've configured, here is a is a 508 00:30:04,620 --> 00:30:09,930 happy little is a happy Helm chart for you to, to use. And 509 00:30:09,930 --> 00:30:12,990 this is this is the baseline of what we'll take this, these are 510 00:30:12,990 --> 00:30:14,640 the guardrails around that application 511 00:30:15,030 --> 00:30:16,890 John Osborne: openshift templates didn't really have all 512 00:30:16,890 --> 00:30:19,770 that layering approach. It just had one master template with 513 00:30:19,830 --> 00:30:24,150 environment variables. And and it could it could do some random 514 00:30:24,150 --> 00:30:26,820 generation of numbers and passwords and things like that. 515 00:30:26,820 --> 00:30:30,120 But it wasn't, it wouldn't layer on all your changes, per se. 516 00:30:30,960 --> 00:30:31,350 Yeah, it 517 00:30:31,350 --> 00:30:33,780 Tariq Islam: wasn't true templates. Yeah, exactly the way 518 00:30:34,050 --> 00:30:36,600 the way Helm is, but I'm thinking just, just 519 00:30:36,600 --> 00:30:37,260 functionally, 520 00:30:37,890 --> 00:30:39,930 Jamie Duncan: I think you certainly have similar feelings. 521 00:30:41,280 --> 00:30:43,320 It's been a while since I used an open ship template. 522 00:30:43,349 --> 00:30:45,779 John Osborne: But yeah, the most ops experience, Jamie, like, do 523 00:30:45,779 --> 00:30:48,449 you think I was touching on something on the assumptions? 524 00:30:48,660 --> 00:30:51,480 Jamie Duncan: Yeah, no, yeah. And I think those assumptions 525 00:30:51,480 --> 00:30:55,020 like, you know, if you're facing inward, if you're building an 526 00:30:55,020 --> 00:30:58,440 internal application, you know, if you're the ops team in a 527 00:30:58,440 --> 00:31:01,800 DevOps environment, those relationships are very different 528 00:31:01,800 --> 00:31:04,230 than if you're sending your code out to an end user. 529 00:31:05,220 --> 00:31:07,680 John Osborne: Or if you're a large enterprise that has never 530 00:31:07,680 --> 00:31:10,860 met, the I've seen, never met them before, regardless of 531 00:31:10,860 --> 00:31:12,600 Jamie Duncan: what's on your paycheck, you know, those end 532 00:31:12,600 --> 00:31:16,710 users are completely different experiences, depending on you 533 00:31:16,710 --> 00:31:20,040 know, how those things are structured. I think it's really 534 00:31:20,040 --> 00:31:22,980 interesting idea, and I've always thought of them as sort 535 00:31:22,980 --> 00:31:27,570 of a linear, like an evolutionary path. And they may 536 00:31:27,570 --> 00:31:28,260 not be, 537 00:31:29,490 --> 00:31:31,200 John Osborne: I'm gonna be in the shower, and something's 538 00:31:31,200 --> 00:31:36,000 gonna hit me. I'm gonna really set articulate this. So yeah, it 539 00:31:36,000 --> 00:31:41,880 does feel like a handoff. Almost like a ticket driven handoff 540 00:31:41,880 --> 00:31:44,100 versus an API driven handoff or something. 541 00:31:44,129 --> 00:31:46,229 Jamie Duncan: Yeah, that's a good way to put it. Yeah. Yeah. 542 00:31:46,709 --> 00:31:49,409 One is not less valid than the other one. They certainly have 543 00:31:49,409 --> 00:31:50,819 different use cases. Mm hmm. 544 00:31:51,419 --> 00:31:53,759 John Osborne: And they both exist, and aren't going 545 00:31:53,759 --> 00:31:54,149 anywhere. 546 00:31:54,240 --> 00:31:56,490 Jamie Duncan: Yes, as a neither. God knows neither going 547 00:31:56,490 --> 00:31:58,290 anywhere, anytime soon. Yeah. 548 00:31:59,159 --> 00:32:01,139 Tariq Islam: Yeah, I know, we got to wrap this up. But I just 549 00:32:01,139 --> 00:32:05,609 I would love to see. I would love to see if anyone out there 550 00:32:05,609 --> 00:32:11,159 has actually leveraged both, right? We've got plenty of 551 00:32:11,159 --> 00:32:14,819 examples of folks leveraging one or the other. But I'd love to 552 00:32:14,819 --> 00:32:18,299 see a success story around. Yes, we we've we've got the use 553 00:32:18,299 --> 00:32:22,439 cases. And we've we've leveraged both across, you know, a few 554 00:32:22,439 --> 00:32:27,029 different processes between cicd between in, you know, 555 00:32:27,029 --> 00:32:30,119 infrastructures code and the whole Git ops thing that we've 556 00:32:30,119 --> 00:32:32,549 been doing. I'm sure there's someone out there. 557 00:32:32,760 --> 00:32:34,530 John Osborne: There's definitely I've seen people online saying 558 00:32:34,530 --> 00:32:36,630 that they were doing it. But I think to me, the cognitive 559 00:32:36,630 --> 00:32:40,050 overhead is now just too high. Yeah, of what you're doing. 560 00:32:41,339 --> 00:32:43,679 Jamie Duncan: You know, we talked a lot and that, you know, 561 00:32:43,679 --> 00:32:47,549 Episode 10, we'd have our first guest. If someone does want to 562 00:32:47,549 --> 00:32:51,959 weigh in there, whether you have sort of expertise level taking 563 00:32:51,959 --> 00:32:54,689 something in production into production with both of these 564 00:32:54,689 --> 00:32:59,909 tools and where you're using them. Let us know. And come hang 565 00:32:59,909 --> 00:33:03,119 out and talk for a few minutes about make us a little smarter. 566 00:33:05,250 --> 00:33:07,800 John Osborne: Always down for always down, I get smarter. 567 00:33:09,630 --> 00:33:11,370 Jamie Duncan: It rarely works for me, but yeah, it really 568 00:33:11,370 --> 00:33:15,720 works. So I think this is it's kind of we again, we were 569 00:33:15,720 --> 00:33:19,140 talking about this before we started recording. I think we're 570 00:33:19,140 --> 00:33:22,980 going to do the next few episodes on sort of CDW 571 00:33:22,980 --> 00:33:27,300 principles, sort of ci CD concepts and where it all lives 572 00:33:27,300 --> 00:33:30,780 in the Kubernetes ecosystem right now. This was for a few 573 00:33:30,780 --> 00:33:34,350 minutes. We talked about this being the The first of those, 574 00:33:34,380 --> 00:33:39,450 but we've picked it, we jumped right into the middle. So I 575 00:33:39,450 --> 00:33:41,430 think we'll start that in earnest here in a couple of 576 00:33:41,430 --> 00:33:45,780 weeks. But is there anything else we want to we want to kick 577 00:33:45,780 --> 00:33:48,000 this? Do we want to kick this from another angle? Before we 578 00:33:48,000 --> 00:33:48,690 wrap up? 579 00:33:48,839 --> 00:33:51,569 Tariq Islam: I feel like we've arrived to a point here where we 580 00:33:51,569 --> 00:33:55,019 could we could confidently tell folks that there are use cases 581 00:33:55,019 --> 00:33:55,589 for both. 582 00:33:55,770 --> 00:33:57,390 Jamie Duncan: Yeah, I think we've certainly teased out a 583 00:33:57,390 --> 00:34:01,230 pattern that none of us had fully recognized before. We'd 584 00:34:01,230 --> 00:34:05,700 all have pieces of it, but I've never had the whole fault 585 00:34:05,970 --> 00:34:08,100 Tariq Islam: and tools why it's why I love this podcast with you 586 00:34:08,100 --> 00:34:13,320 guys. It's always a journey of discovery and reflection. That 587 00:34:13,470 --> 00:34:14,040 fantastic tastic 588 00:34:14,069 --> 00:34:19,739 John Osborne: Yep. There's about 400 projects in the CNC f 589 00:34:19,739 --> 00:34:23,219 landscape for CIC D some I definitely think we should pick 590 00:34:23,219 --> 00:34:27,509 some of them that are that we think are gaining traction or 591 00:34:27,509 --> 00:34:29,159 resonating the most with customers. 592 00:34:30,360 --> 00:34:33,930 Tariq Islam: See ICD is probably going to cover about 380 of 593 00:34:33,930 --> 00:34:38,490 those. So I mean, we'll get we'll get right to it in the 594 00:34:38,490 --> 00:34:39,300 next episode. 595 00:34:40,200 --> 00:34:42,900 John Osborne: Do you want to you want to tackle DevOps to without 596 00:34:42,900 --> 00:34:44,040 being too hand wavy? 597 00:34:44,670 --> 00:34:46,590 Tariq Islam: Well, I just causing me some anxiety over 598 00:34:46,590 --> 00:34:46,770 here. 599 00:34:48,630 --> 00:34:53,070 Jamie Duncan: You have a stay in pain behind my left eye. Sweet 600 00:34:53,100 --> 00:34:55,290 Tarik, you want to bring us you want to close this out? I mean, 601 00:34:55,290 --> 00:34:57,810 this was this was this is the topic you wanted to cover? 602 00:34:58,260 --> 00:35:01,320 Tariq Islam: Yeah. No, we we So yeah, just to wrap it up 603 00:35:01,320 --> 00:35:06,060 covered, we covered helm. And its history in, in this space 604 00:35:06,060 --> 00:35:10,620 that goes way back, I mean, way, way back, covered its use cases 605 00:35:10,620 --> 00:35:14,820 kind of its architecture and the iterations in between. We looked 606 00:35:14,820 --> 00:35:18,720 at, at customize and we we juxtapose them, we compare and 607 00:35:18,720 --> 00:35:23,250 contrast and them. And I think the coolest thing about this 608 00:35:23,250 --> 00:35:26,520 episode is I think we've all kind of arrived at a place where 609 00:35:26,520 --> 00:35:29,640 we feel like there's definitely room for both, but it's got to 610 00:35:29,640 --> 00:35:32,940 be really well thought out. So again, if there's if there's 611 00:35:32,940 --> 00:35:36,540 anyone out there that's done this in any real way, 612 00:35:36,900 --> 00:35:39,240 Jamie Duncan: we'd love to hear from you. Very cool. Thanks for 613 00:35:39,240 --> 00:35:41,460 joining us on episode 10 of K files. 614 00:35:41,820 --> 00:35:43,140 John Osborne: Thanks, everyone. Stay safe. 615 00:35:44,790 --> 00:35:45,390 Tariq Islam: Thank you.