Today on Day Two Cloud we engage in strategic thinking about cloud, workload repatriation, costs and spending, DevOps and Kubernetes, and more with guest Tim Banks. Tim is a Lead Developer Advocate at Dell Technologies and has done stints at AWS, Equinix, and the Duckbill Group. Besides being a techie, Tim Banks is also a Brazilian Jiu Jitsu champion and world-ranked competitor in his class.
We discuss:
- The realities of cloud repatriation and where repatraited workloads might end up
- Whether we’ve updated our tooling to support cloud ops outside of public cloud
- Whether Kubernetes is a piece of that tooling puzzle
- What cloud changes mean for the people building and operating software
- DevOps vs. Platform Engineering
- What cloud looks like in 2023 and where it might be heading
- How to counter an armbar (Just kidding, that’s for a different podcast)
- More
Show Links:
@elchefe – Tim Banks on Twitter
Transcript:
[00:00:09.930] – NedWelcome to Day Two Cloud, and today we have a special guest on it’s, tim Banks. He’s a lead developer advocate at Dell Technologies and he’s also a Brazilian Jiu Jitsu master. Top ranked in the world. It was pretty interesting conversation, and I really appreciate that he led with that. Ethan yeah, that was fun. [00:00:30.480] – Ethan
So if you get lost in the jiu jitsu, conversation only last three minutes or so, but it was super interesting. And it leads in very nicely to the overarching theme of this discussion, which was strategy. We have a lot of thinking we do about cloud and cloud strategy, whether you should be bringing workloads back on premises and if you do, what that looks like and how it impacts finances and all of that. It was all too short of a podcast. Ned I feel like we could have gone another hour or two with Tim. He had so many insights. [00:01:00.930] – Ned
Yeah, but the main takeaway for me was really just, like, strategic thinking and having a plan and understanding the larger context in picture. So enjoy this conversation with Tim Banks, lead developer advocate at Dell Technologies. Tim Banks. Welcome to day two. Cloud. Super pumped that you could join us. Why don’t you tell us a little bit about yourself and your history in tech? Because I know you’ve worked for many interesting companies like AWS, Equinix and the Duckbill Group, and I’m certain that’s given you a unique perspective in the industry. [00:01:36.750] – Tim
Yeah, it has. But I’ll start off with some people ask me, Tell me about yourself, and a lot of people will lead off about being, like, a tech nerd or something like that. I will not do that. I am the five time consecutive and reigning Pan American Brazilian Jiu Jitsu champion in my division. I am the two time American national winner and a three time winner of the silver medal in Brazilian jujitsu in my division, as well as holding an international ranking of number two in the world. [00:02:07.270] – Ethan
Now, Jiu Jitsu has come up in a few different podcasts. I’ve listened to different folks that have engaged in that. I’m trying to remember who and when I’ve heard that before. But it’s not karate and it’s not several other disciplines. Can you give us the what’s the quick compartmentalization of jiujitsu as a practice? [00:02:22.740] – Tim
So there are folks who have T shirts that I love that call it Mexican Ground Karate, and it’s not exactly true. Brazilian jiu jitsu is a grappling martial art based on throws like you see in judo, as well as positional play like you see in wrestling, as well as joint locks and chokes as submissions. It is based on mostly leverage and controlling distance, both on standing up and on the ground. Jujitsu is just as dangerous off your back as it is up top or standing. There’s no real striking. Well, I mean, there’s not what you would most people consider striking, but if I go to an ankle trip on you very aggressively, it’s going to feel like you got kicked. But it is a sport that is originally designed to neutralize differences between size and strength. And there is no ideal jujitsu body, which I think is fantastic. Like, if you can have a very, very long reach, and maybe you can do well with that. But I, who five foot eight and built like a bowling ball with corgi limbs, will still manage to get around people who have quote unquote, ideal athletic bodies. [00:03:42.610] – Ned
So, you know, yeah, a good friend of mine practiced Brazilian jiu jitsu for a while, and she did not have what you would call an ideal body type for what you’d think of a typical martial artist. But she had a great time with it, and I think it really played to her strengths of being a little more grounded and able to grapple with people. So that’s very cool. I’m curious what drew you to that, because I find a lot of the times our professional choices also bubble up in our personal choices of what we pursue in terms of things to do outside of work. [00:04:13.570] – Tim
So I think it’s interesting. Jujitsu is something I’ve had an interest in since I saw UFC One back in 1993. We were all discussing about how we are of a certain age. I remember seeing UFC One as an 18 year old back in 1993, which I thought was very interesting because there was this person, very small, you know, looked very much out of his out of his element compared to these big, strong grapplers and, you know, strikers and stuff. He saw other sports, and it didn’t matter. He just wrapped them up, chucked them out, every single one. And I thought that was amazing, that not only could he minimize damage, but he could also submit them all like that. I thought that was intense. I thought it was pretty well. And there was never a situation where he looked like he was out of control. And I really admired that. And so I feel like one of the things that might have appealed to me about it is that nature of the very strategic nature of it. I’m funneling you into places where I know exactly what you can and can’t do, and therefore, all your moves are predictable, which is really people who practice you at a high level can do. [00:05:26.970] – Tim
And if you don’t practice at their level and you see them do that to someone, or you’re the one having done it to you, it’s nightmarish. Because nothing you do is the right thing. Every decision you make is bad, and you go from bad to worse very quickly. [00:05:40.110] – Ned
You’re just recounting my teenage years right back. It sounds like you should take up a career in infosec more than anything else. [00:05:49.870] – Tim
I won’t take up a career in infosec because I don’t have a problem clearly stating my identity on the Internet. And I feel like in order for me to infosec question and I don’t look really great in baseball caps or I don’t like where hoodie is in my head over because my hair is fantastic. So I don’t know that I’d be a fit for infosec in general. [00:06:09.370] – Ned
Ethan and I were discussing your hair when you dropped off for a moment, and it is absolutely fantastic. [00:06:14.380] – Ethan
I totally jealous, man. My hair’s left me. [00:06:17.610] – Tim
I have nothing left. But no, I suppose I could be in a lot of roles in tech, and the skill sets that I have as far as when it goes to strategic thinking and identifying where I want to go with something have helped me out a lot in my career, especially engineering and architecture, especially reverse architecture. Or consulting with folks to try and get them to the place they want to be, or trying to save them money on their bills, et cetera, et cetera. That really helped. But the thing that I think is also really important is that the thing that I found that really helps me know AWS jiujitsu is both being taught and teaching via conversations. Jujitsu is not something where you just do this, you do this, you do this, you do this, right? It’s very partner based. There are not a lot of static drills or katas or forms in jiujitsu. Everything is based off a reaction to your partner or your opponent. Right? [00:07:21.340] – Tim
So if you step a certain way, I can do these things, but if your foot is not there or your weight is shifted differently, or your hand position is here versus there, I can’t do anything I wanted to do. Right? [00:07:34.560] – Tim
So it’s a lot of following and figuring out what people are doing, understanding what they’re trying to do, and relate that into what you’re trying to do, and almost a way of having a conversation with somebody. So if you take words away and you make it about movement and balance and position, jujitsu is very much a conversation, right? It’s a debate. It’s chess with dire physical consequences. [00:08:02.730] – Ned
I feel like that should be printed on the T shirt somewhere. Maybe it already is. [00:08:06.810] – Tim
I don’t know. I’m sure Twitter bot has already scraped this by now and advertising it. [00:08:13.610] – Ned
It’s entirely possible. [00:08:15.310] – Tim
I forgot. Wait, no, there’s no bots on Twitter because Elon Musk got rid of them all. He did. [00:08:19.070] – Ned
I forgot about that. [00:08:20.210] – Tim
Well done, Gigi. Bro. Gigi. [00:08:24.610] – Ned
Well, it’s interesting that you bring up the conversational aspect of technology, because I’ve worked in a consulting capacity before, and the worst consulting engagements I ever had were where people were not having a conversation across the table. It was the consultant saying, this is the way we do things, and this is how we’re going to do it. Not listening to their customer and the customer going, well, this is the way we do things, and that’s how we’re going to do it. And not listening to the consultant. And as a result, the engagement just went extremely poorly every time. Instead of having that conversation that you’re talking about where you’re learning from each other and coming to a good plan. [00:09:00.560] – Tim
In the end, nothing you do for someone else is going to be good without context, right? So if you’re building an application, if you’re delivering a consulting engagement, if you’re delivering an infrastructure, if you’re baking a cake, if you’re building a house, it doesn’t matter what you do, if you don’t have context with the person who’s consuming that, you’re masturbating. You’re doing something one way. You don’t care about the results, you’re telling you’re not listening. [00:09:27.260] – Speaker 4
Right? [00:09:27.720] – Tim
And that is really great for it’s like karaoke, right? It’s for the performer, not for the audience. Right. But when you’re delivering something to somebody, if someone has a need and they’re coming to you for a service or product or a good, you have to for how they’re trying to use it. Otherwise you’re going to build something that’s useless. [00:09:44.190] – Ethan
Tim, one of the things you mentioned along the way here was that you help people save, have saved money on their cloud bills. I’m guessing that’s part of your history with the duck bill group. How big of a thing is that still? Is that still a chief focus, a need that people are looking for? [00:10:03.020] – Tim
It is huge. It is probably the number one concern. And I’ll tell you, right, and we’re all going to remember when the cloud first started became a thing. And what did they tell us? Oh, you’ll save so much money because you’re not going to be spending it all at once. You’ll treat it like a utility, and then you can just pay every month for what you need and you don’t pay for what you don’t need. As if that was a problem with the data center and data center, what they really were saying was like, well, data center, you have to predict your capacity, right. You have to have forecasting, and you have to make these purchases ahead of time. But then you have so much unused stuff or you don’t have enough stuff, right? And that was like a promise of cloud, right? And you just kind of forget about the bill. But the bill for enterprises became huge and out of control, and every time you wanted to do something else, you paid more, which is not the thing you saw in data centers, right. You paid for it one time, right. And you didn’t have it, but you had a power that was it may. [00:11:06.790] – Ethan
Not have had the most efficient use of it because you were designing for peak load a lot of the times. And so, yeah, there was unused capacity laying around, but right. It was a capital expenditure. You paid for it once and then you used it until it was unusable. [00:11:19.690] – Tim
And then so what ended up happening is customers got bigger and bigger and bigger. Their bills got bigger and bigger and bigger. And so now they had to rearchitect their infrastructure and applications for cost savings. And if they didn’t, they just paid big bills, right? They had to architect for stateless spot. They had to architect for how they’re going to do logging and observability. They had to architect for what does redundancy and resilience look like that were different than they would have before because cost was a concern, right? They had to prevent you from using certain instance types. They had to take away credit cards. And so the cloud providers tell you, well, if you want to save money, spend more, right? They’re like, why don’t you turn this opex back into a capex and commit for three to five years for $50 to $500 million a year to save money? Reserve an instance, right? Buy an instance ahead of time for this amount of time, although you’re not actually reserving capacity, right? Because you may reserve an instance, still not be able to spin it up, but we want you to pay for this instance ahead of time so you get savings. [00:12:29.930] – Tim
And once customers started doing that and they were like, oh yeah, sure, we’ll give you we’ll commit to spending this much either ahead of time or before the time for however many years. What was the advantage of going to the cloud versus having it in the data center, right? [00:12:46.100] – Ned
And on top of that, at least when you had the data center, you had an asset that you could depreciate against. You buy reserved instances. As far as I know, the auditors and accountants have not approved that as a capital expense. You can depreciate against. [00:13:00.410] – Tim
Well, you can go through this thing of reselling a reserved incident, but it’s not like you can lease out and get spot for it. Maybe you can, I don’t know, but not like you can just, all right, cool, we’re done with these. We’re going to sell these servers on the open market and get the money back for it. But the thing is, though, again, you only pay for them one time. You don’t continue to keep paying for them. After a while, you get your money back if you compare it to the cost of a cloud. And they’re like, oh, well, you don’t have to pay for data center folks. And I’m like, well, they don’t. You got to pay them every month. But it’s way less than my AWS bill. [00:13:35.990] – Speaker 4
Yeah. [00:13:36.390] – Ned
And if you were coming from a data center model, it’s not like all those people went away, you know, I hope not. And your cloud engineers on top of it, or you learn both and burn them out. [00:13:48.950] – Tim
And the other thing that happened is now that we’re on cloud, we don’t need Ops. So you had like, no ops and all this stuff like that. You’re turning people that know how to code, and you’re trying to turn them into infrastructure engineers, and you get some of the things that we have seen since then, which are not necessarily great, and then AWS goes like, oh, well, we’ll pay you even more. So you have to learn even less or take care of even less. But the end result was you still have to pay more. Now, some companies, very small companies, agile, that are very agile and don’t have right now the bandwidth, I will say, to recruit infrastructure engineers is very efficient. It helps out a lot, but once you go over a certain size, it doesn’t help. And I say bandwidth because I don’t say cost because it’s not like you’re not spending money, you’re just giving it to AWS. [00:14:39.270] – Ethan
So Timothy is you’re having these cost savings discussions, or did have, and you’re finding that Gosh cloud is all of a sudden starting to look like the data center model. Has that been driving cloud repatriation or has that been driving we need to re architect our applications so that they run more efficiently in cloud and we can save money that way. [00:14:59.230] – Tim
There’s been a few things that have been driving cloud repatriation, right? There’s been compliance issues. For some industries, it’s still more practical to own your own assets than to lease them or put them in a place where other people would have access to. And that’s just for box checking. I don’t think that’s the best case, honestly. But it is a driver, right? Cost is a very big one because sometimes you really cannot rearchitect something for cost without significantly changing how you operate to the point where sometimes it’s easier just to lift and shift into the data center. I’ll be very honest, the most normal use case for repatriation isn’t a total repatriation, it’s a hybrid model, right? I’m going to take everything thing that I reserved for three years or that I would reserve for three years. I know what kind of CPUs they are going to be, I know what my capacity is going to be, and I’m just going to literally buy that hardware and run in the data center, right? And that’s the most efficient because there are some things that you really can’t repatriate very well. There are some things that don’t work well that are not suited for data centers, stuff like that. [00:16:10.630] – Tim
Gigantic pools of large data analysis and stuff like that. Data centers haven’t really answered the multi petabyte object storage thing at low cost like you can do on cloud, right? But I will tell you this, if you’re pumping out a lot of bandwidth to the internet, you’re going to do it way cheaper in a data center than you ever are going to be at any cloud provider. AWS, Google and Azure are all with like a half a cent per gig of each other on their list prices. And I can tell you from behind the scenes they’re not too far off together. In their private pricing prices either, but they’re still more going to be more expensive than you’re going to pay for out of a data center. And that’s by design, right? It’s free to get in, cost to get out, right? I don’t know what else that sounds like to you, but that’s a business model that has been used exploitatively before. And so if you do run large amounts of data being pushed out, especially in like media or something like that, showing media, if you can have enough storage in your data center, it makes a lot of sense to run it through there or any other kind of thing where you’re going to be putting in a lot of data. [00:17:27.060] – Tim
It makes a lot of sense. You’ll pay for yourself much more quickly than you would pretty much with any other use case. So there is a good hybrid architecture that’s basically based on a lot of known compute in a data center or in a private cloud that puts out a lot of data to consumers of wherever they may be. And then you push other storage, like longer term storage for analytics later on down the road into the cloud where the storage is cheap, the inbound network is cheap, and then you can spend your money doing analytics there on the back end, right? [00:18:09.210] – Ned
I think of some companies that went all in on the cloud native promises and started taking advantage of the specialized services in the clouds. They’re going to have a real issue migrating any of those applications back on prem because the same constructs don’t exist. There’s no lambda on prem, at least unless you want to rent AWS Outposts for another monthly fee. [00:18:34.050] – Tim
No, please God. AWS Outpost is the fact that AWS Outposts exists, the fact that Eks anywhere exists, the fact that Google Anthos exists tells you that even the cloud providers know that there is desire and will for a lot of companies to move back on premise for various reasons and they’re still trying to capture that revenue. But you’re right to the extent, and I will say this is a huge gaping failure of the cloud native community is that you can run Kubernetes kind of really anywhere. And that’s one of the things that you’re really seeing is it’s becoming ubiquitous almost, but serverless functions has not really been, there’s no standardized way to do it. So you get lambda in one place, you get functions for other cloud providers, but there is no I can just run the same library in a data center. I would probably say that’s probably by design the various cloud companies have gone in on containers a lot. Not the cloud native technologies, but serverless just kind of went the way of the Dodo as far as it comes to having a good open source solution for running serverless functions. There’s a couple there, but nothing with the kind of adoption that you would really want to see that makes it an industry standard. [00:19:55.090] – Tim
I would like to see that fixed. There are a lot of things I would like to see as far as being able to schedule a workload literally anywhere, whether it’s a cloud provider, whether it’s in your private, private data center. Now, folks will say that’s Kubernetes and I would have agreed with you a couple of years back, but now Kubernetes are getting treated like VMware VMs were ten years ago, and the fact they’re gigantic and they require a lot of overhead, and they usually require a vendor or a lot of expenses like that. And that’s not the way Kubernetes was designed. It’s just what Kubernetes? [00:20:27.900] – Ethan
What do you mean, overhead? You mean, like, just to get a cluster, a Kubernetes cluster off the ground that can start hosting containers? It’s a lot of overhead to run. [00:20:36.170] – Tim
Kubernetes in production at scale is a lot of overhead just for the control plane, just for the logging, just for the observability, just for the scheduling, just for all these other things. And I’m not saying that that hasn’t been necessary for Kubernetes, but what Kubernetes does now is not what Kubernetes was designed for. [00:20:58.370] – Ethan
What are you getting at here? [00:21:00.530] – Tim
If you think about Borg, when it first started and some of the things that Kubernetes was doing, it was doing very kind of lightweight stateless workloads. Micro services that were coming out and SaaS, hey, I was going to do this and then die. I was going to do this and just die. Just do this and die. A lot like serverless was being done now, right? Or I’m going to maintain a resilient load, a resilient pool of workers for this stateless application. But as folks started moving state onto containers, which was supposed to be ephemeral, once you start preserving state, you can really no longer be truly ephemeral. [00:21:37.730] – Speaker 4
Right. [00:21:38.180] – Ethan
State has to say femoral state is an oxymoron. Yeah. [00:21:43.810] – Tim
That’S just how it goes. Like, as soon as you will start saving state, but still run this. Okay, well, now Kubernetes becomes different. It’s not a container really anymore. I mean, functionally it is I would just say by how you call it, everything it is. But functionally, it’s a VM now. [00:21:59.510] – Speaker 4
Right. [00:22:01.030] – Tim
And you have to maintain people want to know insights inside, too. They want to know how it runs. They want all these other information that you wouldn’t really care about so much if it was really a truly stateless, small, lightweight container. [00:22:13.620] – Speaker 4
Right. [00:22:14.080] – Ethan
If you do enough, it doesn’t work and it dies. Why do you die having on this highly instrumented container with all the durability? [00:22:20.500] – Speaker 4
Yeah. Right. [00:22:21.480] – Tim
And so we’re laying those on top of cloud based instances, which we weren’t supposed to care about either, that we now care about. [00:22:28.690] – Speaker 4
Right. [00:22:30.190] – Tim
And we’ll just say the most inning thing I still see today is people who run a JVM inside a container, inside a VM, inside a hypervisor, just all these layers of abstraction for different things. Sounds like I remember when I just used to be able to log onto a server, run a job, and then go, right. And it didn’t have a lot of overhead. And I would like to see that again, some approximation thereof. I know that there’s folks working on that. Like, the project that comes to mind very quickly for me is Aura, which is done by Chris Nova and some other folks. They’re starting to turn Ranches on it now, which is just it’s a runtime. You spin it up, you schedule jobs on it runs on the bare metal server, you run jobs on it, and then those jobs go away. Right. I think it’s pit one or PID zero. [00:23:20.640] – Ethan
Everything old is new again. [00:23:23.490] – Tim
But you think about it, like, what if we layered on top of layer on top of layer on top of layer on top of layer? We have done this in the name not of simplicity, but changing where the complexity lies. We’re trying to take away complexity from the developer, but we didn’t do that. We’re trying to make our engineering teams less complex, but we didn’t do that. We’re trying to make our internal tooling less complex, but we didn’t do that either. [00:23:54.410] – Speaker 4
Right. [00:23:55.690] – Tim
So we have very complex systems even to do simple things still now. And the notion of logging onto a server via Ssh running a thing is gone now, saying we should go back to that, but I think there are some things we could have brought back to that where let me put it like this. Do you remember the first time you played Quake Online? [00:24:23.250] – Ethan
Oh, yeah. [00:24:24.320] – Speaker 4
Oh, yeah. [00:24:25.240] – Tim
The first time I played Quake Online was on a 56K dial up modem, right. 32 people in the server. What do you think those servers were running back then? [00:24:37.430] – Ned
Oh, jeez, was it like 286 maybe? [00:24:40.600] – Tim
No, they were probably running like I ran a Quick server on a dual Penny and Pro 180 with like half a gig of memory, something like that. [00:24:49.050] – Speaker 4
Right. [00:24:49.680] – Tim
Running on dialogue because the code was so efficient. The gameplay was great. I still miss a gameplay from regular Quick. Gameplay was fantastic and it wasn’t visually stunning. [00:24:59.220] – Speaker 4
Right. [00:24:59.660] – Tim
But the experience was still fantastic. [00:25:03.170] – Speaker 4
Right. [00:25:04.150] – Tim
And I don’t like the way that we have mutated and evolved what the user experience is AWS to what we have now. [00:25:19.750] – Ned
Do you think the biggest problem here is the excess capacity that people have available to them? [00:25:26.300] – Tim
Oh, absolutely. We become reckless of storage as we become reckless with amount of storage within a CPU memory, because we have more of it, we use more of it. [00:25:34.270] – Speaker 4
Right. [00:25:34.700] – Ned
Efficiency is unnecessary because just because of the nature of Moore’s Law, the doubling of the number of transistors on the processor, the expansion of memory, and then the expansion of storage. And we had that at the data center level and then virtualization rolled in. It’s like, well, now you can create 50 VMs per, you know, physical server. And then the cloud rolled along and said, forget about those bounds. It’s boundless. You have effectively infinite compute storage and everything else. Have at it. And what we did is we built these abstractions on top of abstractions because we wanted to make it convenient and. [00:26:11.340] – Tim
Simple and we didn’t care about how much it cost. We didn’t care about what resources were involved. Like, if you think about it from a sustainability model, cloud computing is the worst thing it could have ever happened. [00:26:23.150] – Speaker 4
Right. [00:26:24.250] – Tim
Because if you think about it, if we had a data center and we had to buy ahead of time, we know exactly what our power consumption is, we know exactly what our heat output is going to be. We know exactly what all these other factors because we had to know ahead of time when we bought it. Now we don’t care, right? And then if we do care, we’re going to trust Amazon to tell us how much is coming out of there. We’re going to trust Google to say, oh no, we’re carbon neutral. [00:26:49.080] – Speaker 4
Right? [00:26:50.610] – Tim
Trust Microsoft to tell us, no. Yeah, hey, this is great. Sustainability is awesome. Why have we abdicated our own responsibilities to cloud companies? [00:27:02.870] – Speaker 4
Right? [00:27:03.320] – Ethan
Well, it’s at the point now where new data centers are going up and there isn’t enough power in the area to feed a new data center. And so the data center stand up project is in conjunction with some sort of power generation that has to happen at the same time. [00:27:18.670] – Tim
And I look at places like, I remember Sao Paulo region for AWS is still, I believe, some of the most expensive area you have there because it’s so difficult. [00:27:29.070] – Speaker 4
Right. [00:27:29.940] – Tim
And so they’re putting more data centers in an area that’s already having problems with sustainability. Not to say that we shouldn’t have cloud computing presence in Brazil and in South America, but like, can we do something better than just creating infinite capacity in a place that we’re already clear cutting forests? And it may sound like I’m a bit of a tree hugger, and to be sure, I’m not necessarily a quote unquote tree hugger, but I do think we should have an understanding and have a good realization of the impact that we have and then we can make good data to driven choices around that. [00:28:11.070] – Speaker 4
Right? [00:28:11.500] – Tim
We should be intentional with what we do and with cloud computing. We have become reckless because of the. [00:28:19.120] – Ned
Recent changes in the economy. Do you feel that people are going to start really caring more about costs and efficiency and maybe as they move workloads back to a more bounded area that efficiency is going to rise in terms of applications and deployments? [00:28:38.550] – Tim
I think it could, but it’s going to depend. And here’s why. It is very easy to lay people off, right? People do it. You can just. Make stroke a pan. Lawyers have to get involved, but really, that’s about it, right? And so they say, oh, well, we overhired. I think what the sea levels are saying now, we overhired. We’re cutting dead weight, or something like that. [00:29:00.420] – Speaker 4
Right. [00:29:01.950] – Tim
Okay. We’ll accept that at face value right now because we’re just going to play like we’re naive. But you notice they don’t reduce the services. They try to optimize. And I’ve seen this because it’s when I get called in. They try to optimize what they’re doing, but they don’t reimagine what the user experience should look like. Think about this. Think about this. [00:29:24.870] – Speaker 4
All right? [00:29:25.830] – Tim
How many sites or how many services do you have that you log into that have your email address, that have your credit cards, that have your home address, that have your demographic data? [00:29:38.200] – Speaker 4
Right. [00:29:38.420] – Tim
How many different copies of that are on different places all over the world, right. Probably a lot. Why do they all need the same information? Why can’t that information be just in one place and these places read it when you need it, and then it’s gotten rid of when you’re done. [00:29:58.320] – Speaker 4
Right. [00:29:58.860] – Tim
What do you think that would do for what storage looks like, what state looks like, and what the overall impact of what that looks like is? Right. We have AI. All the new advancements to AI were based on incredible amounts of data that were readily accessible to people that we just literally have laying around and have no control over because we’ve given it away in the essence of, quote, unquote user experience. [00:30:25.210] – Speaker 4
Yeah. [00:30:25.550] – Ned
And privacy is definitely an issue there beyond everything else, the unbounded problems. And I like the did initiative. It seems very interesting to me to have a dedicated, distributed Identifier that’s unique to you and you have control over. [00:30:43.010] – Tim
Absolutely. [00:30:43.950] – Ned
That’s very cool. But I want to turn this back to the cloud and the cost question in terms of optimizing code and optimizing applications. I know you’re at Dell now, and you were at Equinix before that, so you were seeing what might be more the repatriation portion of things, where people are, I’m going to buy some servers and use those servers. Do you see people being more careful about what they’re ordering and their capacity and maybe trying to push back on folks who are trying to over provision? [00:31:20.650] – Tim
Some folks are. I mean, we’re very honest, but a lot of times what they’re doing is they’re just trying to take the capex that they put in a cloud provider and put it in a less expensive capex in their own data center, which is a good idea to do from a financial standpoint. But what really has to happen is, like I said, a reimagining of how you deliver an experience, how you deliver experience for the developer, how you deliver an experience for your customer. Right. The biggest hurdle to repatriation is delivering a cloud like experience in the private cloud. And now you can do that with Open Shift and some other Kubernetes distros. You can do that if you god forbid we’re using VMware, the data center, and as well as on a cloud provider like you can with the shipment of VMs around there. But how do you interact with assets and resources in a data center? [00:32:13.070] – Speaker 4
Right? [00:32:14.010] – Tim
And it’s not the same as it is in the cloud. And Cloud has delivered a great experience for that, and hardware manufacturers and private cloud folks are catching up. [00:32:24.450] – Speaker 4
Right. [00:32:25.550] – Tim
But I think as that becomes more the case where you can deliver a cloud like experience to your operations teams and to your developers, the lift of moving into the data center becomes much more of a procurement issue than an engineering issue. Because right now it’s both. [00:32:43.890] – Ned
Right? We got used to the cloud model of operations, which I really like someone who interacts with infrastructure AWS code a lot and does some DevOps processes kind of things. I really like that workflow, and it feels like it could be extremely efficient. And then I look at the data center and people using a remote desktop to manage servers, and I’m like, I would love to see that transformed. And it seems like the tooling isn’t there yet, and I don’t know. [00:33:13.630] – Tim
Okay, well, let me ask you a question, then. What was the last time you worked with assets in a data center? [00:33:19.230] – Ned
That’s part of it is I’m also very far removed from the data center world. So I’m curious, as close as boots on the ground as you are, what are you seeing in the modern data center in terms of tooling and processes? [00:33:33.950] – Tim
So there was a company called Packet, which was bought by Equinix, became Equinix Metal. They had a provisioning tool called Tinkerbell, which is open source. It’s in CNCF now. And it is for provisioning bare metal. [00:33:49.810] – Speaker 4
Right. [00:33:50.660] – Tim
Well, we use the Equinix Metal was a variation thereof, little secret sauce to provision bare metal instances with whatever environment variables you wanted to using. Going through Tinkerbell, you can do it through terraform, you can do other things like that. [00:34:05.220] – Speaker 4
Right. [00:34:05.540] – Tim
So infrastructure code does exist for the data center assets. Dell we have CSM CSI models or CSM CSI modules that will allow you to provision container storage directly to our storage devices without needing an intermediary. Or they can also work with Open Shift, I think, VMware, Tanzu Rancher and a few others, right? So that you can provision storage and compute to some extent without having to go in remote. You can treat it like infrastructure as code. The Idrac. You can still interact with a remote desktop, but you can also interact with the API. And anybody who’s ever working with data center knows that Idrack is a lifesaver. [00:34:50.250] – Speaker 4
Right? [00:34:51.710] – Tim
It’s like Dell Recovery Assistance Console or something like that. But there’s also like, telemetry around not only your resource usage, but about sustainability. And power output, heat output and stuff like that, all that information can come back to you directly using Otel. So data centers have been working with the open source community to deliver a more cloud led experience, both for infrastructure provisioning, but also observability and observability where it matters for sustainability, not just capacity planning. We’re working on things where you can figure out cost and bill back and charge back and everything like that. The gap is not as large as you think it is, but again, people have to get out of their own heads on what the experience is going to be like because there’s a consumption model challenge here. [00:35:47.950] – Ethan
Because if everybody’s used to the AWS API and that’s how they’re going to consume cloud and then you bring it back on premises and try to do you don’t have the AWS API, you have some kind of an API or series of APIs perhaps, depending on what you’re trying to provision and what tooling you’re using. But it isn’t the same. There’s going to be some sort of operational impact. [00:36:07.750] – Tim
No, and honestly, if I could waive a wand, I would have an omnibus universal provisioning API that could work with anything, right? And people have tried to build that for sale, right? But if there was at least just an open API standard that everyone could like, okay, well, our API interacts with this, and these calls get into that. And we’re used to working with APIs. If AWS can create 748 services to launch a container that all interoperate with the same API, I’m sure the rest of the industry can do the same. But in the end, right, you have to dedicate the resources to it and you have to tell people it’s important. And I think that’s the point. That’s one of the main things that I’m looking to do. And the folks in the devrell team that we have at Dell are looking to do is like, what resources can we put in and what help can we give the industry to figure out how we want to deliver a cloud experience, or how we want to deliver a provisioning experience that is smooth and easy for everybody? Because if you’re a cloud formation shop, it’s not like you can take it anywhere, right? [00:37:26.270] – Tim
But if you use ansible Terraform or Palumi, et cetera, any of these other kind of infrastructure code tools that are not proprietary, you’ve got options, right? I think Terraform is probably the one that’s the most widely used and most widely usable. And so maybe that’s the standard as an example. But people create their own modules that interact with Terraform. It’s just writing an API that interacts with an API that’s a good model for use. And so I don’t see why we can’t do that for a lot more stuff or just have that be the standard, whether it’s provisioning containers, whether it’s provisioning hardware, where it’s provisioning storage. So there’s a lot of potential for folks to be able to deliver a universal experience. We just need to people to realize first that’s important and then actually pick. [00:38:22.190] – Ned
Up and do it and train the folks who are going to be delivering on that promise. [00:38:28.610] – Tim
If people can figure out how to provision stuff in Kubernetes, they can figure out anything. That’s true. [00:38:34.710] – Ned
Just need the support from the managerial structure to let them get the training or give them the time to do it. [00:38:42.230] – Tim
Exactly. That’s why to Ethan’s point, right, when people are talking about how they’re going to cost cut or what it looks like in this industry, that becomes very important because manager can look like if we were to do this, that gives us more options, more options for saving money, more options for making money. [00:39:07.150] – Speaker 4
Right. [00:39:07.490] – Ned
But that’s a strategic outlook. And we know you like strategy. I do and I like strategy. But sometimes the people that the finance department, they’re not always about the strategy, they’re about the next quarter. [00:39:22.510] – Tim
Sometimes it is very weird because if you work because I’ve worked extensively with finance finance departments finance departments are usually more strategic than anybody else in the company because they are used to forecasting and prediction as a way of life. And how this little thing that happened now will affect all those forecasts and predictions down the road. Engineers could do better by listening to finance and understanding that model. Because finance is all about predictability and handling and resilience, right? Because think about your finance department has to be able to understand and react to swings in demand, in cost and income and things like that. They have to be able to make your company resilient to it. And if things need to change as far as how much they’re putting out, right, they’re going to know what they can do that’s still going to be sustainable and what’s not going to be sustainable. And so it’s interesting that a lot of times engineer fights with finance, but engineers and finance should be best friends. [00:40:35.450] – Ned
Well, go make friends with your finance person. I like it. [00:40:39.290] – Tim
It’s true, right? I will go so far as to say that finance can be better friends and engineers and sales are. But if we really want to get some people upset that’s the tea really. [00:40:51.810] – Ned
Want to stir the pot. [00:40:52.680] – Tim
Yeah. [00:40:57.890] – Ned
One thing I’ve been hearing, and you’ve probably heard this too, getting back to the process idea. I’ve been hearing DevOps is dead. And that platform engineering is the way, the truth and the light. And I just wanted to get your reaction and your thoughts on that, especially in lieu of the conversation we’re having where we need people to learn new ways of doing things and maybe shrink down some of the abstractions we’ve been getting comfortable with. [00:41:24.810] – Tim
People who say DevOps is dead never have an understanding of what DevOps is in the first place. Cut and dry plain and simple. You can ship that sound bite and it will be true. [00:41:33.940] – Speaker 4
Right? [00:41:34.720] – Tim
DevOps is not dead. DevOps is now the default. As a culture, as a practice. DevOps is now the default. It’s not a tool, it is not a person, it is not a department. It is a way you operate. DevOps includes platform engineers and platform engineering. [00:41:54.610] – Speaker 4
Right? [00:41:55.480] – Tim
But to say that DevOps as dead and platform engineers are the default is just somebody who’s trying to sell a book or get on a talk, it’s nonsense. [00:42:05.780] – Speaker 4
Right? [00:42:07.050] – Tim
Platform engineering is exceedingly important. It’s an important skill set, it’s an important kind of role, but it still falls within the realm of successful DevOps culture to have those just like Sre, just like software engineers, just like pretty much anything that goes in between via that you whiteboard and an application that a customer ingests. [00:42:39.730] – Ned
So I guess sort of round out the conversation. We’ve talked about a lot of different things. You mentioned layoffs. We know layoffs have happened recently, and there’s definitely an opportunity there as companies look to maybe repatriate some of their applications on prem. So if I’ve been laid off, let’s say, and I’m looking for my next gig, what do you see as the core areas to focus on if you’re trying to pick up a sustainable set of skills and a new job? [00:43:10.170] – Tim
I think that as we go into the future here, I don’t feel like folks who specialize in a specific engineering realm are going to be as hirable AWS others. And I say that to mean that people who have been in shops where devs did everything, if you’re frontend dev or back end dev, or people have all these arguments on React or JS or whatever, right. Those specific technologies can go away. To be that specialized or to be that invested in this one thing is not a resilient kind of career choice. [00:43:53.060] – Speaker 4
Right. [00:43:55.210] – Tim
Maybe it’s cobalt. [00:43:56.640] – Speaker 4
Right? [00:43:56.960] – Tim
Because that’s not going away, apparently ever, fortunately. But JavaScript react angular. I don’t understand the framework wars myself, but picking up and understanding, actually understanding infrastructure, actually understanding what happens on the other side when you commit something to GitHub, right. What actually really happens, understanding how the sausage is made and how it’s delivered and being able to make that happen anywhere is really going to help you out, if only because you are now more hireable by companies that you may have to wear more hats at. [00:44:36.790] – Speaker 4
Right. [00:44:37.430] – Tim
But also because you’ll have a better understanding of what’s going on and you’ll have context, so you can contribute at a much higher level to companies where they have to run a tighter ship financially. Because you understand the impact of what’s happening when you hit commit, you understand what those servers are going to do. You understand how much data that’s going to take. You understand the cost of that. [00:45:00.750] – Ethan
That is hard to package on a resume, Tim. That is difficult to market yourself that you have those skills unless you have the right human looking at your resume as you tell that story, that explains your broad skill set and your understanding and your depth and your background so that they get it. I know how I’m going to plug this person in. What happens so often with modern resumes, it feels like it’s all about keyword searches. It’s all about, do you have the certification? Do you have this particular skill set? Are you terraform certified? Et cetera? And so I think people that are hunting for work are caught in this difficult position where if I’m the hiring manager, I want the person you just described. If I’m on the other side looking for work, I feel like I’ve got to have all these specific skills certifications on my resume or I’m not going to get noticed. [00:45:54.110] – Tim
I think it’s interesting because you talk about the resume. My resume, I think, now has reached five pages long, and I will not cut it down for anybody or anything, right? But the most important thing on my resume is a suburb of my skills. The beginning, the description, the objective, right? The narrative parts of my resume. That’s what’s important. Cover letter, maybe. But when you read my resume, the first thing you’re going to read is my name. And then I’m going to give you my little short bio of what I’m done. I’m going to tell you that I have a wide set of skills and that I understand this and I understand that. I go from the top to bottom, right? It’s a good, healthy paragraph. I think it’s actually two paragraphs. Use those narratives to talk about what you are and what you are, what you can do, and then use a resume to back that up. And I think what happens is and not what happens, it’s a very real thing. We in technology, in the tech community, as a group of engineers are just awful storytellers. It’s true, right? When you look at user stories, especially project managers have to be storytellers. [00:46:59.150] – Tim
They have to be product managers have to be storytellers. They know, have to know how to make stories. They have to know how to listen to stories. And if we cannot tell the story, we’re going to have a hard time getting people to understand what it is we’re trying to do. And so people who have that full understanding, or at least a fuller understanding, I should say, of what all is entailed with the process of creating something and then delivering it are usually going to be your better storytellers. Now, storytelling is skill, right? You can be a rack and tour. You can be very comfortable with sitting and talking about things, and that’s one way to tell the story, but it’s not the only way to tell the story. [00:47:38.180] – Speaker 4
Right? [00:47:39.030] – Tim
But I would much rather have read someone who’s got two, three paragraphs like that than going somebody to GitHub who’s all solid green. [00:47:46.390] – Ned
That just shows me, you know, how to game the GitHub algorithm. [00:47:49.420] – Tim
That’s it, that just tells you it tells you you commit a lot of code. That doesn’t mean you write it well. [00:47:54.340] – Ned
It doesn’t mean it’s good at all. I really like how you kind of tied it all in there because we started out talking about understanding, context, having a conversation. And I think part of the hiring process is having that conversation and knowing how to tell your story and listen to the other side and what they truly need from a prospective employee and being able to speak to that. And maybe it’s not a fit and being okay with that, that’s another big thing, is don’t shove yourself into a particular shape just because that’s the shape of what they need at the time. Not everybody has that luxury. And I want to acknowledge that sometimes you just need a job, but assuming you’ve got a little bit of cushion and you can actually search around for the job that fits, you find one that actually fits. [00:48:40.790] – Tim
Let me explain about that too, because I hear a lot of discourse about, well, if you make it too long, then people aren’t going to read it, or people aren’t going to do that, or people going to do this. I’ve got all these resumes to go through, blah, blah. If that is your disposition, you’re just so busy, then don’t often do the interviews. Have someone else who’s not lazy, who actually cares about people and cares about finding a good candidate. You have them do the interviews, right, because the point where we’re just lazy, we’re going to look for what we’re comfortable with, what we’re used to looking for, and that’s going to be good enough, leads you to really, really crappy outcomes. And I think it’s an awful way to do it is to cater to the whims of the arbitrary whims of an individual accounting manager. Like if you are not invested in looking for talent and looking for people who can really do good work, then don’t do it. [00:49:28.870] – Ned
Right. And I think that’s a great point to ride out on. Tim, if people want to hear more from you, and I think they might, where can they find you on the interwebs? Where do you have a presence? [00:49:40.320] – Tim
The best place to find me is on Twitter. I’m at El. Chef El C-H-E-F-E on Twitter. Stay tuned and look at developer Dell.com. It’s going to go undergo its own little reinvention here, and we’re going to be turning it into a place that you’re going to want to spend more time on and you’ll be able to see a lot more of my content there as well. [00:50:04.350] – Ned
Awesome. Well, Tim Banks, lead developer advocate from Dell Technologies, thank you so much for being a guest today on day two. [00:50:10.290] – Tim
Cloud, thank you all too. I appreciate it. Absolutely. [00:50:13.490] – Ned
And hey, thank you dear listener, for listening to the whole episode. Virtual highs fives to you for tuning in. If you’ve got suggestions for future shows, we would love to hear about them. You can hit either of us up on Twitter. We monitor Day Two Cloud show, or you can fill out the request on our website. Day Twocloud IO hey, it’s a new year and Day Two Cloud is looking for new sponsors. If you’ve got a way cool Cloud product and you want to share it with our audience of It professionals, you can become just such a Day Two Cloud sponsor. You’ll reach several thousand listeners, all of whom have problems to solve. Maybe your product fixes their problem, but they’ll never know unless you tell them about your amazing solution. You can find out more at packet pushers net sponsorship and until next time, just remember cloud is what happens while It is making other plans.
Podcast: Play in new window | Download