87% Cheaper Than AWS? This Move Is Changing the Game

Max: 00:06 This is a conversation with Byron Dill, who’s a principal solutions architect at Devft. Devft is a rebrand of a combination of 2 companies, Server Central, who’s been in business for over 20 years doing dedicated servers and data centers, and Turing Group, who was, an AWS focused consultancy. And this conversation, what I like about it is it really focuses on what companies are seeing in the real world on how to decide where to put things. Do you keep things in AWS or in a hyperscale cloud, public cloud? Do you keep things in a data center?

Do you find some kind of middle ground? What are you dealing with? And how do you make your life better and easier? You know, obviously, Chris introduced us and, you know, said, hey. We should we should we should chat.

And he gave me a little bit of your backstory, but, you know, I I mean, other than, like, what I’ve I’ve seen in writing, it’d be good to kinda get a little bit of, you know, straight from the horse’s mouth, so to speak.

Byron: 00:54 Byron Dillon, principal solution architect here at Daft. Been with the company about three and a half years. The 1st 3 years, I was, on the delivery side, on the technical teams, helping run those organizations. You know, an opportunity came up to kind of do what I found I kinda was enjoying more, than organizational management was, you know, really talking with clients about their problems, and and their challenges. And I think a lot of that is I didn’t grow up in the MSP world.

Before joining Devd, I was actually a client of Dev’s. And, you know, buying these services from them and understanding what the drivers are on the other side of the table. You know, I’ve been in IT for about 25 years, bouncing all over the place. Started at, University of Missouri in in academics, you know, cutting my teeth on big AIXSP frames. Before that, some definitely some NT and Novell NetWare for, for the folks that remember that one.

Max: 01:42 I miss Novell. I’ll just, you know, just put that out there.

Byron: 01:45 It was elegant, but nowadays, it’s it’s look back and you’re like, poof, why did we ever put up with that? But, you know, from there, some pharmaceutical work, couple years in entertainment, which was tons of fun, health care start up in there.

Max: 01:58 Oh, come on. You’re not gonna name drop her Oprah? Like, you know

Byron: 02:02 Yeah. I can I can name drop Oprah? That was honestly some exciting and really fun times. We were revamping the website, and when you’re revamping a website that gets tens of millions of hits a month, it’s definitely challenging. You know, it’s if you you wanna talk about burst, capacity.

And, you know, this was in 2006. This was you know, there really wasn’t a cloud or auto scaling, and so you really had to have, you know, the static load, capability, but then the ability to handle the onslaught. You know, the show would run at 3 PM CST, and you could sit there and watch the the dashboards at 4 o’clock. Or if, you know, haven’t haven’t helped us, as she said, go to the website right now and look at x. You know, it would just the the graphs would just be exponential.

It was it was pretty insane. You know, when you’re doing days where, you know, you’re 50,000,000 hits in a day, it it gets pretty intense.

Max: 02:56 For the people that weren’t around 2,006, I mean, that is, like, a lot.

Byron: 02:58 It was a lot.

Max: 02:59 A lot of it.

Byron: 03:00 And, we tried to do a or we we pulled off a a live streaming, back in I think it was 2,008 to early 2009 where she actually streamed some of her stuff live on on on the web. And we we pushed some CDNs to nearly the breaking point, with, you know I think it it wasn’t until Michael Jackson came along that they they kind of surpassed, some of the the online live streaming, and, that was fun. You know? I learned a a ton there. It was interesting working in a nonregulated industry, just the different work mode, working with extremely talented and creative people there.

Max: 03:39 So, hopefully, somebody was watching the taping and gave you the heads up versus after the taping. I was like, oh, by the way so, yeah, this is happening, and, you better be ready for it in the next 3 hours.

Byron: 03:49 Yeah. Well, everybody had TV on their desk, so we were actually able to watch the the recordings going on. And there were a few a few moments of you could almost hear people fall out of chairs sometimes with you know, we got a lot of work to do in, like, 30 minutes to do it. But, no, it was it was a it was a phenomenally fun time, in my life. But yeah.

And then, you know, health care start up, I did, about 10 years with a, alternative asset management company, so financial technologies, dealing with that on a global, scale. And, you know, eventually, really wanted to come and work at an MSP where daily I get to talk to those kind of clients.

Max: 04:26 You know, the cost delta of of now, you know, terminology is repatriating workloads from the cloud. AKA, you move to the cloud, you realize maybe it wasn’t a great idea. You move off the cloud, you know, back into a data center. And my rant was something along the lines of, like, we’ve done this a bunch of times now. Enough that, you know, you can you can model this pretty effectively.

And, our average cost reduction coming out of an AWS environment scale is 87%. So purchasing the equipment, signing signing contracts, getting colo, getting net cross connects. I mean, they’re like soup to nuts when you when you look at it and you say, okay. After you invest all this money and you figure out what your, you know, leasing or capital depreciation schedule is or whatever thing is, and you look at that versus AWS. You’re like, okay.

That’s an 87% reduction from AWS. And I’ve been in lots of conversations where, you know, when that number comes up, you know, the responses is okay. You’re you’re lying to me. And it’s like, no. No.

No. I’m actually not lying to you. This is just how much AWS costs. And and there’s cost optimization strategies with AWS and, you know, EDPs and RIs and savings plans and rearchitecture and all this different stuff. But, you know, even doing all that stuff, I I think it’s shocking for people when they find out how much the cloud actually really cost them.

Right? I I look at this in a couple of different ways. And I’m and I’m curious to see what you’re seeing, you know, with Daft and and how you help people through this. And and and Daft has a has a client. It’s very famous.

You know, Basecamp talks about this very publicly. And, DHH is a is a is a very strong personality on the Internet. And, of course, Basecamp just announced also that they are repatriating as much as they possibly can out of cloud. And and he kinda got into the cost structure a little bit for them. And and some of their other staff went into more details what what this really means.

But, you know, I come out of the world, you know, same with you. Started with NT and Novell and a s 400. And so, you know, data centers were like just what we did. And and then data centers went out of vogue. And now it’s all cloud.

And now companies are kinda dipping their toes in data centers a little bit again. Kind of a long lead up here into in in into what I wanna talk about, which is, you know, end of q42022 and beginning of q12023, you know, I feel like there was still a lot of loose money out there. And and and now I’m having conversations with companies. It’s like, okay. We have an OKR for, like, this year of reducing our spend by 50%.

Like like, no joke. We have to figure out how to cut 50% out of our budget. You know? So if I came to you and I said, okay. Hey, Byron.

I have this problem. You know? Like, what do what do I do here?

Byron: 06:48 Again, it’s it’s, for me, it’s starting with, you know, what what talk to me about your journey that led you to the cloud in the first place. You know? What what what led you to there? Was it cost savings? Was it, you know, getting out of the colo or running your own data center?

Were there challenges with the pandemic. I mean, we saw a lot of just operational best effort, in 2020 of people just trying to figure it out. So that’s usually where I I kind of wanna start with folks. And, again, it’s it’s starting at that business level and working down to what got them there is gonna give a lot of information about where they wanna go. And and in some ways, it’s almost like inception.

You’re you’re really trying to get them to tell you their story about what they need, and then you’re crafting solutions around it.

Max: 07:30 I’ve been through phases of, you know, you had to run PEA WebLogic with Oracle database on Sun Microsystems if you were a VC funded entity. Otherwise, like, it was just you were weird. And it wasn’t until companies like E toys came out and and went public with the, oh, we’re running a mod Perl stack with, NetApp filers and MySQL back end, and we have this huge infrastructure and website. And our cost differential is so massive that people were like, oh, wait. It’s okay if you don’t run BEA.

And I I feel like in cloud, there were some, like, nontechnical forces, or you say operational best effort. You know, I like that terminology. But you couldn’t pick up a management magazine or book for a span of a decade that didn’t have some, you know, digital transformation, cloud transformation, like this pressure of just, oh, you’re weird if you’re not migrating everything into the cloud. And and I and, you know, how much pressure did that create? And was that, you know I mean, our company’s unwinding this now.

Byron: 08:22 I think they’re coming to the realization of of kind of a I wouldn’t say a rush to the cloud, but I think there was pressure from, you know, some of the marketing, some of the, the the the documentation that was out there. You know, like, you mentioned it, you know, save a ton of money moving to the cloud. Reduce your you know, the pressure on your teams. And I think it in many ways, people maybe saw it as a panacea. You know?

It’s it’s the tool that will solve everything. It is the Leatherman toolkit of everything you’ll ever need for IT. And I think the realization is that’s not the case. It is it is a unique tool that has the ability to solve problems for you, but it doesn’t mean you throw away your old toolbox. And that’s really getting to that kind of hybrid cloud conversation that I like to talk about.

But definitely seeing the the shift from everybody’s in the cloud. I hear everybody’s in the cloud. We need to be in the cloud. You know? And and I wouldn’t say it’s lemmings, but at times, it was, you know, what do we wanna do in the cloud?

And and I was part of some of those conversations, you know, about 7 or 8 years ago where it was, what does it cost to move everything to the cloud? Like, well, if you wanna get there in 6 months, you’re talking a forklift, you know, and it’s going to cost x because you’re really just replatforming. You’re not doing any kind of rearchitecture. Part of my argument back then was we have lofty ambitions of getting into the cloud and then rearchitecting in situ and taking advantage of all the cool tools that they have. Realistically, that never happens.

And I think that’s where a lot of companies are are finding themselves with kind of legacy workloads, legacy ways of operating with not taking advantage of the cloud. And that’s where we’re getting this kind of not a full repatriation, but just more that that hybridity that they’re opening up to, you know, realizing there’s really cool tools in the cloud, but there’s also really compelling arguments for having, you know, on prem managed services. You know, the the thing that we we kinda laugh here about is, you know, the we’ve been doing hybrid clouds forever. I wouldn’t say forever, but for quite a quite a long time. You know, we have an AWS team that we have in house.

They don’t compete, with with our colo teams. They they augment, and both teams really allow us to bring that hybrid approach, to select that best tool. You know, it’s, our VP of revenue, mister Biggs here, Aaron Biggs, you know, talks about high you know, depth is was hybrid before hybrid was cool.

Max: 10:37 You know, I’ve I’ve I’ve seen a lot of people shocked when they actually start modeling out that cost. You know, we we wanna do something different here, and it and it feels like, you know, a lot of projects become DOA very quickly when they start actually quantifying, you know, what is this gonna take for you know, and whether that’s multi cloud, which I don’t really believe really exists. You know, I see multi workloads in different clouds. But I don’t I don’t really you know, what what marketing defines as multi cloud, I don’t I don’t think exists out in the world. But, like, take marketing defines as multi cloud, I don’t I don’t think exists out in the world.

But, like, take relatively benign things like, you know, CDN. There are much better CDN options for you in the market than what the cloud platform provides, But you still have to egress off your storage platform to that other CDN and then deliver from that CDN. And then when you factor in the cost of that egress from that storage platform, unless you have a really high hit rate, it’s you gotta factor that cost into the whole thing. I mean, it it drastically changes the financial equation of this.

Byron: 11:31 They they definitely want you to stay in their ecosystem. My tagline here when, you know, definitely when I’m talking to clients is, you know, the the major cloud platforms have figured out how to quantify and assign a value, a dollar value, everything you can do on their platforms. Just like you you said, they make it very easy to get in, difficult at times to get out, but that’s not impossible. A lot of times, those are conversations about what what really has to come out, what really has to to egress, you know, and and and what’s the the strategy behind that so you don’t incur some of those costs. But it it’s challenging because you can say, hey.

You can repatriate all these things, but we we wanna be careful that we don’t just say, hey. You’ve saved a bunch of money by, you know, maybe repatriating some things, but now your DX or ExpressRoute is ballooning in cost for you. And if you don’t kind of look at that in that larger picture, you’re gonna find yourself right back in the same position of you’ve traded one cost for another. You’ve optimized something, but now another piece isn’t optimized the way you expected. And you talked about the workloads.

That’s that’s where it could become as critical. Mapping those workloads, what’s tightly coupled, what’s loosely coupled, where are you going to have those potential fault lines that you have to address? And I think that’s where I and with this repatriation efforts that DEFT is working with clients, it’s taking the time and care to do that in a in a way that doesn’t, you know, just see the pendulums shift back the other way and set the client up for in 3 years, they’ve got a whole new, like, wait. Why are we on prem? We need to be in the cloud.

And now you’ve you set up this oscillation that, honestly, no one has time for. You know? Every every tech team that I talk to, they’re they are working flat out. They’re probably seeing, like you said, their budget’s cut. There’s an OKR.

We have to reduce spend by 30, 50%, and and they’re under some immense pressures. And last thing you wanna do is set them up for a quick win that’s gonna cause a bunch of work.

Max: 13:17 I feel like there’s, you know, the the there is a push into serverless, and there’s there’s some interesting platforms come out. Right? You know, I mean, Google Firebase is interesting and Fly. Io and all these different things. You know, there’s some really interesting, you know, like, serverless native kind of environments that, you know, like, akin to what Heroku started with.

I it feels like the majority of companies aren’t really using these things. Like like, Lambda is being used, but, like, the percentage of of deployment in Lambda versus, you know, EC 2 versus, you know, a a kubernet a a a Docker, you know, a Kubernetes based platform. It does feel for me at least that most companies have invested in in containers, not so much other things. And then, thankfully, those containers are relatively portable. I mean, there’s a you know, it’s not like, oh, you can just move a container.

I mean, there’s a lot of other, you know, stuff that you have to factor into it. But so the auto scaling group fires off e c two instances that then loads a single container per instance. So each instance is just a single container, and that container is what the application is. And it runs up, and it was like, oh, we don’t know how to really actually, like, push and manage code to our fleet, so we’ll just make a container. And you get into this conversation, you’re like, okay.

That’s interesting. That’s probably not what you wanna be doing here. You know? Like, there’s there’s better ways to approach this.

Byron: 14:22 Yeah. And it’s that’s kind of a traditional approach to the new technology. I think a lot of times, it’s it’s the having the time to truly understand the new tool, get the get the right get the training to leverage, you know, AWS, Azure, GCP in a way their tool set can bring value to what you’re doing. When I talk to, you know, clients here at definitely EC 2 and s 3. You know?

They’re they’re computing their storage. And I think part of that is kind of a paradigm shift in the way they need to think about their infrastructure. It’s not a this or that. It’s where are my workloads best suited for it. And containers, you know, they’re great.

I don’t see a lot of companies that struggle with containers. While you may pick up portability, in being able to move workloads around a little bit, there’s always a a opportunity cost there. Usually, that comes in complexity, and orchestration is great. Kubernetes is amazing, amazingly complex, and it’s approach it as it’s it’s a new tool that you can use to solve problems, not a we can put everything that we have taken maybe out of colo, and now we have it in AWS or or, you know, instances. Now we need to containerize it.

That’s not necessarily the only path. It’s not a one way street. And there are ways of of parking workloads where they are most effective. Containers are great, but I’ve I struggle to work with clients that truly understand what can be containerized, what the orchestration layer looks like on top of that. Just because you can doesn’t mean you should.

You know, you may have Docker containers and it’s working great for you. That doesn’t mean you have to move into, you know, an orchestration layer like Kubernetes or or adding complexity to that environment is not the answer.

Max: 16:03 If you’re running e c 2 and s 3 and EBS volumes, spend the time to go through and and figure out if you have unattached EBS volumes. I mean, the amount of money I see being spent are just wasted. And it and that’s a really simple win. Like, if you’ve got any sort of auto scaling legacy if you’ve been in if you’ve been in any cloud platform for a long time and there’s something like EBS on that platform, spend the time to actually go through and figure out if you have unattached EBS volumes. I mean, it’s it’s it’s amazing how much money gets sucked up there.

Byron: 16:29 Absolutely. And that’s due diligence that everyone should do. I mean and and a lot of these platforms have frameworks you can use and tooling you can use, CloudWatch. There’s all sorts of, you know, cost spend analytics tools out there, and that’s that’s part of the new management if you’re gonna be in these hypervisors, is doing that cost analysis. What is your behavioral delta?

Because you usually don’t find out until the bills hit, and you’re gonna owe it. You’re gonna owe it. And it it’s it’s on you to, to figure out how you can control that. And and here at DAFT, you know, we we we run, with our AWS team, we run all architected reviews with our clients, regularly, even clients that are in great shape. You know?

We work with them to say, hey. You know? Part of our QBRs should be leveraging this framework that’s that’s out there. It’s published, and it’s pretty easy to run. And that’ll identify not just places where there may be, you know, volumes or instances that are oversized or we can take advantage of new opportunities, but it’s us taking the time to, you know, bring that hybrid approach that DEFT brings to say, hey.

This workload that used to be very elastic and you you could take advantage of all those is really stabilized, and maybe it’s it’s a legacy workload now. Maybe now that’s time to transition that, repatriate that piece so you can focus on, you know, the next gen of what you’re trying to build for your company that’s able to take advantage of, you know, all the the fantastic tools that AWS, and the other hyperscalers bring to the table.

Max: 17:59 You, you had said something that I think we should we should talk about a little bit more, which is, you know, there’s more than there’s more ways to run containers in Kubernetes. Kubernetes is a very heavy amount of infrastructure that might it it could be appropriate or it could not be appropriate. And there’s a lot of alternatives to running Kubernetes, which, you know, I mean, become very simple based on what you actually need. I mean, there’s very simplistic ways to deploy containers. I mean, how much of this when you’re engaged with a client and they’re asking about operational efficiencies or operational support or cost optimization.

I mean, there’s lots of different, probably, levers or or questions that come in more pain points, probably. Let’s use pain points. Right? What they come in with. But these things also require you know, the the white elephant in the room is they require engineering time and engineering work.

You know, looking at us at a problem and saying, hey. We’ve got this issue that we wanna resolve, you know, and and saying you know, knowing that, like, we can solve this problem, but but that’s just it. The white elephant is it’s gonna require their engineering teams to do work. Like, how I mean, what what is that journey like for you? And how do you help clients understand, you know, the the trade offs of, like, do we go down this path or do we not?

You know? And and knowing and I I think the frustrating part for me also is, with companies looking for cost reductions, you know, the easiest lever for most to pull is is layoffs. It’s frustrating knowing that there’s infrastructure costs that can be tackled that just you know, that seemingly unreachable because there’s just not, you you know, the ability to actually go and deal with it from an engineering time standpoint. So, you know, what do you what do you seeing? What do you see with that?

Like, how do you deal with this, and how do you help people with that?

Byron: 19:35 For our AWS team specifically, that goes back to kind of those well architected reviews. Starting there, coming up with recommendations, take a little more holistic approach. And then, you know, the sky’s the limit here at Daft. You know? We we we can deliver on that promise of technology and but it’s also asking the question, you know, not just can we do this, but should we do this?

And getting folks to kind of take that moment of, is this right the right path that we wanna embark on? And the teams may not have the bandwidth, the expertise, and that’s where here at you know, I talk to our clients about, you know, we’re not gonna have a solution and and sell you just that solution. We’re gonna partner with you, because we know that the thing that we put in place today with your teams and and dev’s teams, to solve that challenge, that’s gonna shift. I mean, the moment we deploy it, the needs for your firm are gonna shift, the needs of the company, and and you’re gonna start to see, I wouldn’t say config drift, but you’re gonna see a a an a drift in the objective. And, you know, that’s why you have, you know, the OKRs are yearly, not, you know, hey.

We’re gonna have one OKR to rule them all. I mean, the moment you put that out there, new data is gonna flow in, and we’re gonna have to be agile. We’re gonna have to shift. And having somebody that, that really helps you deploy those solutions, and understands it at a very low level, can help you with that. And it’s not just deploy it and come talk to us again in 6 months or when the next thing blows up, and we’ll help you then.

It’s staying engaged with those teams, knowing that they’re under that pressure of couple people maybe got let go or we’re under you know, we’re not sure what we’re supposed to do here. We know they’re they’re trying to save money, and I think everybody’s seeing what’s going on in in the news right now with with lots of layoffs out there. And I think it is the easy way to to pull that lever.

Max: 21:12 Nothing in tech that’s temporary is actually temporary. I mean, this this goes from, like, as basic as, you know, patch cords. Right? There’s, like, no such thing as a temporary patch job, you know, patch cable in your in your in your computer room. But the horror stories and there’s a certain amount of paralysis as well.

You you know, and I’ve I’ve wondered about this for a long time, and I’ve been trying to figure out, like, the the root cause of it. But, I mean, here you know, an example I’ve I’ve is relevant to this. I know a company, and they started going through a a cost reduction exercise with their AWS environment, you know, big spend, you know, 8 8 figures plus a year, you know, had a significant spend in Dynamo. Right? So, you know, it’s one of those things that just kinda gets it just grows.

And what they found was, you know, after digging into it, you know, they had a table in dynamo that was very large that didn’t have TTL set when the table was created. And you you know exactly where I’m going.

Byron: 22:05 Growing growing.

Max: 22:06 Right? And so then what and they’re like, okay. Great. Let’s purge out this table. Well, then they find that, like, okay.

In order to actually purge the table, they have to go back and do an update operation against the table, which is gonna cost a lot of money. It’s not like it just you go and you do a delete star from Dynamo and you move on. Right? Like, it’s it’s a it was it was significantly expensive. And this company internally, you discussed, debated, argued, whatever terminology you wanna use.

I mean, 6, 7, 8 months. You know what to do about this Dynamo table? And in the process of doing that, spent 1,000,000 of dollars on Dynamo trying to come up with a decision of what to do with it. And I think part of that’s, you know, like, nobody wanted to be to own the solution or or take responsibility for it. I mean, there’s, like, a certain amount of, like, keep your head down, you know, kind of that comes out of it.

You know, what they had to do eventually, and they realized was the best thing to do was they created a new table effectively in Dynamo, and they, you know, they wrote to both. You know, 13 months of data, they deleted the original Dynamo, and it was the cheapest you know, know, what they decided was the cheapest way to do it. But then, you know, less than a year later, went through and laid off 30% of its engineering staff. How do companies manage this healthy? Like, how do you help people get into a healthy place of, like, dealing with infrastructure at scale and, you know, the implications of these decisions?

Byron: 23:24 For me, I I I think a lot of it is is coming from that financial industry is is talking about risk. It is a very common thread that you can talk, and it’s it’s the risk of doing something, and it’s the risk of not doing something. And if you can quantify that for your business, dollars and cents, resources, time that it’s gonna take to do something, you can really boil things down. And and at the end of the day, you know, companies are in business to make money. And I do think that at times, IT teams are seen as a cost.

They’re not an enabler of a company, but they’re seen as a cost, cost code on a sheet somewhere. And I think it’s important for those teams to also understand the business language, and we work with clients like that. Like I like I mentioned earlier, here at Deft, we I’ll talk to a CFO, but they’ll have technical people on the team. So you’re kind of threading the needle of weaving stories about, you know, business objectives and technical objectives and financial and risk and telling compelling stories. But it’s really getting the technical teams to understand it’s it’s not just servers for servers.

And at some point, my favorite discussion is to what is your company in business to do. Unless you’re an specifically an IT company, it’s probably not having a large IT team. It’s probably not having a ton of infrastructure sitting around. We work with clients constantly that are on business objectives, and they need to free up time to focus on those without the kind of the legacy pieces of their infrastructure kinda weighing them down. So there’s always a risk of you you you wanna move to Azure.

You wanna move to the cloud and save money. What is the opportunity cost there? What is the if you do, if you don’t analysis? And I find a lot of a lot of folks have done the if we do analysis. It’s right there.

That’s usually the first one they go to. The other one is, you know, well, what if we don’t put these workloads in the cloud? What if we, you know, work with DEFT and come up with a hybrid cloud solution that allows them to take not just advantage of of DEFT’s, you know, data centers and and our managed services, but our brainpower that we have here. One of the more compelling things here is this first place I’ve worked with, it’s it’s just geeks everywhere. You know, there’s incredibly smart people from the CEO, through our data center operations teams, engineering support.

And that is a that is a resource that it’s hard to put a dollar value on that when we’re talking to clients, but it brings a an adjustment to that cost. You know, it’s your team doesn’t have to necessarily do all these things, on their own.

Max: 25:55 We’re talking about compute and public cloud. Moving to the cloud does not save money. Anybody that thinks still that moving to the cloud is saving money, it’s not saving money. It’s solving a different pain point for you. You do not save money moving to cloud.

So, like, if you’re in a conversation right now, anybody listening to this, that, you know, oh, we gotta save money by moving to the cloud. Just know that you’re not gonna save money moving to the cloud. Now there is real cost and there’s real overhead with managing physical infrastructure. You know, if you buy a server and you put any put a server somewhere, you know, what is your life cycle on that piece of equipment? You know, like, if you buy anything, you know, for your company and you and you have an asset on your books, you know, you’re gonna depreciate that asset over some period of time, and then you you have to, like, replenish that asset.

Server equipment and network gear is no different. You know, you’re you’re on a cycle. Like and, you know, it doesn’t necessarily mean that you have to replace it every 2 years or 3 years. You could replace every 4 or 5 or 6. I mean, there’s there’s different ways of of structuring these things.

You know, becomes one of my big arguments of going out and finding a strong stable partner, you know, like, grew up doing this. I mean, how when when did I mean, DEF goes back, you know, how long now as a data center company? I mean, you

Byron: 27:01 know, as far as years. Yeah. We’re we’re well over 2 decades now as a networks and colocation.

Max: 27:06 You know? And this is this is you know, earlier, I said, okay. You can you you know, we see cost reduction of 87% leaving Amazon for your own stuff. But, like, going to a bare metal provider or a managed colo provider or a virtual server provider or an in IaaS provider or whatever the different terminology is, it’s like, okay. You’re not gonna save 87%, but the reductions are still massive, and you don’t have to get into the habit of like, oh, who do we hire to now go, you know, physically touch a server?

So, also, you’re in this conversation. Like, that that’s a solved issue as well. Like, don’t don’t you know, I don’t I don’t we don’t have to harp on this too much, but it’s it’s something that

Byron: 27:37 I I’ve done those cost analysis. You know, the hey. We’re gonna save a ton of money. Let’s get the the spreadsheet out and let’s crunch them. Let’s let’s talk about, you know, that that total cost.

You know, the opportunity costs, the soft dollars are where you really can get bit too. Completely agree. And then I fought that battle a couple times in my career. Now I wouldn’t say here at depth I fight the battle because I definitely have a different perspective as a hybrid provider where we’re able to we’re able to bring the best of breed from from all of those, platforms and solutions to, to to build something that’s gonna work for our clients.

Max: 28:09 I I mean, I don’t even know what the count is today. AWS has a lot of different services. I mean, there is lots of different ways to to fire up a compute instance. There’s lots of different ways to have a container. There’s lots of different ways to have, you know, data analytics or MapReduce.

I mean, it’s you know, there’s and even within those ways, there’s lots of ways of of doing things. And and and this is the other side of it. When you when you look at and you talk to an an organization, you’re like, well, what’s your skill set internally? Can you go out and be an expert in 280 plus AWS services and also have a bench of people so that way you can account for sick time vacations PTO, you know, you don’t need. So how much of your conversation, you know, with a company when you start getting into this dialogue and and discussion touches on in house their internal expertise and capacity of just, oh, we’ve we’re thinking about switching to a column or database.

What does it actually mean we don’t know? But now how do we manage it? Right? Like, how does that conversation go?

Byron: 29:07 A a lot like what you’re describing. The the the number of Lego bricks that these hyperscale providers put out every year. I mean, AWS, you know, reinvent. We send our engineers and architects out there every year for exactly that. You know, they’re continually learn continuously learning what the new technologies are.

What’s the next iteration of existing technologies? You know, what are the next instances that are out there? You know, what are the what are the ways we can pull different levers that because AWS is responding to, and Google and Azure, responding to their client needs and producing, you know, products that then you can build services and then build solutions on top of that.

Max: 29:46 Wait. Sorry. There there’s another part of this, though. You know, we talk about different cost levers, and you you need to talk about, like, well, architected reviews. Mention EBS, like, go out and figure out if you have unattached EBS volumes.

There’s another thing that that’s really kind of core for me with this, which is, you know, AWS gives you a lot of managed services. Right? You know, you can go out and and, you need a MySQL database. You know, you could just kick off, you know, RDS or maybe you wanna get into Aurora or one of these other platforms. And there’s lots of examples of these things that were that are all based on technology that’s freely available to you in the world.

Right? You know, you know, you can run AWS’s ElastiCache or you can run your own. You can run their version of of Memcached or, Redis or you can run your own. And Amazon charges you a premium to run their managed service. So, Kinesis, you know, really elegant, you know, message bus built on top of Kafka.

And you look and you start talking about here is the amount of money you’re spending on Kinesis versus you could run Kafka on EC 2. And we’re not talking about like a repatriation here. This is still cloud. You know? This is still in AWS.

But instead of running Kinesis, you can run Kafka on EC 2, you know, on our eyes or whatever, you know, a discount plan you you know, you wanna get into. And those cost savings are significant. Yeah. How how much of this? And again, so, like, maybe a company doesn’t have experience running Kafka.

But how much of this gets into a conversation where you say, hey, look, you know, you’re you’re spending, I don’t know, $1,000,000 a month and 200 k of that’s going to Kinesis. And, you know, we can convert that into Kafka and EC 2 and run it for you, and it’s gonna be much less. Like, what you know? You know, how much of that is is also the the, you know, journey that you’re putting people on?

Byron: 31:34 Quite a bit. They come in, you know, like, we talked about at the very beginning. They make it easy to get in, you know, very hotel California. And a lot of those, you know, RDSs, they’re they’re easy to get into. You know, you bring up the the message bus.

You know, I’ve had conversations about that exact thing. Hey. If you spin up a couple of EC 2 instances, you you can run a RabbitMQ on these. And it’s, you know, you’re paying a premium. Getting them to understand that you’re paying a premium for that managed service.

And where is that comfort layer? Because at some point, they may have gone with that because they have no in house expertise there, And that is the opportunity cost there. You know, the if we don’t run RDS, we have to go hire a MySQL administrator or partner with someone that can do that, and there’s a cost there. Making sure you can, like, quantify where some easily wouldn’t say easy. Easy is probably the wrong word.

I don’t think any of this is easy.

Max: 32:24 We’re not talking like a complete and total application overhaul in order to switch from Kinesis to Kafka. Right? Like, it’s still you’re still talking to Kafka.

Byron: 32:30 And that’s that’s what you have to really kind of work with the client to understand, the risks involved. Here, there’s gonna be a savings. You don’t have the turnkey solution easy button. This is how much effort it’s gonna take and, you know, what it’s gonna take for them to either manage that in house or, you know, here at Deft, we’ll we’ll manage that for you. But, you know, we don’t turn around and say, hey.

We’re gonna just turn on all the the RDS and and Kinesis and the managed services in the cloud. You know? We’re gonna build you the solution that’s gonna work, and we’re gonna absorb a lot of that complexity. Their teams don’t have that constant fear of, you know, well, you know, if if if Byron’s not here tomorrow, no one’s gonna know how to turn that one thing on that is critical to, you know, the entire operation.

Max: 33:11 Well and it’s not completely hands off either. I mean, you still have work that you have to do in order to maintain these managed services. It’s not like you can just, like, you know, the the I mean, you know, what’s what’s my, favorite one right now? Managed NAT gateways. Absolute black hole of money sync in a lot of AWS environments.

You can deploy your own NAT gateway in different ways. And, yes, you have to manage it. If you can reduce $1,000,000 a year out of your AWS spend, you can afford to spend some of that savings on either in house engineering or a third part you know, a third party management entity to, like, maintain that for you. I mean, this is like we’re not talking about, like I think that’s what’s so shocking is when you get into this conversation of just how big those deltas become. You know, this isn’t like we’re trying to eke out, you know, 5, 10, 20, you know, 15% here.

We’re talking about, like and, again, engineering has to be involved because things have to change. US average salary is $60,000. When you start talking about engineering, you’re probably in the mid 100. You know, if you’re talking about, you know, reducing spend by $1,000,000 a year. Right?

Like, that’s just jobs. Right? You know, like, I come back to this. It’s like, okay. You had 7 jobs.

Right? So, I’d rather invest that. I mean, companies are people constrained, not technology constrained usually. Like, why wouldn’t you wanna save money on technology to invest in people? Like, that’s it seems like a no brainer to me.

Byron: 34:35 If that goes back to the the conversation we had earlier, I think, though, that the lever that we see being pulled right now is is the inverse of that, which I think is there’s gonna be a price to be paid for that. I I hope it’s not too steep, but, you know, it’s it’s interesting to see opportunities with optimizing technology by using a partner such as Daft that can really, you know, bring some significant cost savings to to companies, allowing them to be more optimized with the staff they have, because you’ve got a a team backing you up, a partner backing you up here at Daft. When, you know, if if you’re reducing staff, there’s always that that tribal knowledge piece, something that’s not documented, something that’s gonna be latent, that’s gonna creep up at the worst time, and and and it’s going to cause a a a true issue for your company.

Max: 35:25 A company like Deft isn’t going to a company and saying, we’re gonna reduce your staff and take over your staff. This is not a conversation that’s going on. I think there’s a big fear still with a lot of people and companies that bringing in an MSP. That’s the conversation that’s happening. And I can tell you from talking with all the executives I have ever engaged with over 20 years, that is not the conversation.

The conversation is I need my people working on stuff that benefits the business directly, and how do I free them up to do that? And that decision for outsourcing or MSP involvement or augmentation or whatever you wanna call it is I have never seen it from a place of, like, oh, I wanna go and do a riff, so I need to get ready for it by hiring this MSP. You know, it’s it’s it’s never that dialogue.

Byron: 36:12 No. I I don’t see that dialogue at all. In fact, it’s it’s my position here with Daft is most companies are not out to run IT shops inherently. So they need someone to help them for exactly what you said. They want their IT teams focusing on what’s gonna drive their business forward.

What’s going to bring in more clients for them, focusing on their business challenges, not worried about the 5 year old server in the data center. That’s stuff that dev can really handle for for our clients. That’s the easy easy pieces of it. It’s it’s freeing those teams up to take care of the business objectives, those OKRs you were talking about. You know, there’s different ways to get there, different ways of enabling that, but they’re not all cost.

You know? There’s growth OKRs as well. Teams have to be they have to have the bandwidth and and, you know, someone that they can call and trust.

Max: 36:57 Something that got lost, I don’t think was really talked about enough at Basecamp and DHH’s post. I think this idea of how you use the cloud effectively, and we get into, like, really into the weeds a little bit on this and how you help people with hybrid. But the point that he made was we launched a new service, and it was a brand new product. It was a brand new service for us. And by the way, we’re I mean, Basecamp is a very mature software development company.

They they have been doing this a long time. They know what they are doing. Right? So we are launching a new service. I launched an email platform called Hey.

And we knew nothing about it, and we didn’t know what we were gonna need. We didn’t know how it scale, and we didn’t know what the response rate would be. And we put that into AWS. And it was a great thing that we did because it was way more successful than we expected, and we scaled way faster than we needed and or that we than we anticipated. And all these different things, we’re able to add resources very quickly because we were in the cloud.

And and I I I really and I I think the the the most powerful thing about this was we took a set a second. We stepped back. We looked at what we needed. We figured out what we needed and where we wanted to run it, and we decided AWS is not the right place to keep this thing long term because it’s just not. And we’re moving it back into a data center.

Or not we’re moving it back into a center. We’re moving it, a a cloud native application. We’re moving a cloud native application into the data center now because it’s a you know, like, that’s where it belongs, and we know what this thing looks like. I’m hoping you’re having that conversation a lot, but I’m not you know, I don’t I don’t feel like that has really penetrated, like, the the consents you know, the the the psyche, you know, out there. The

Byron: 38:25 No. I I think that there’s there’s something to that. You have companies that the cloud is great for rapid prototyping. You don’t have a procurement cycle. You’re able to quickly get the resources that you need, in a way that you don’t have to worry too much about that that scalability or expandability, not not necessarily even elasticity.

You know? Resources are at your fingertips. Where I see, you know, 37 signals, and base camp looking at it is they’ve taken a step back now and and said, hey. Yeah. We we aren’t seeing that explosive growth.

We we understand what’s going on now, and we’ve got an opportunity to take this maybe more static load that they’ve they’ve developed and and and reached and optimize that. But I think a lot of times, it’s that conversation isn’t happening because there’s a lot of inertia, once you get in one place. Like you said, you you patch a cable and you think I’ll come back and I’ll I’ll groom that later. Later never happens. You know, with what Basecamp is doing is they they did take that time to reassess that.

There’s a lot of companies out there that we talked to about, you know, like I said at the very beginning, what got you in the cloud in the first place? You know, what what were the drivers to get you there, and are those still drivers to keeping you there? And when we sit down, it’s, you know, it’s how do we craft that hybrid solution? Hey. You’ve got a couple of workloads that are pretty static.

You know, you can save some money by bringing them back, but maybe if you have merger and acquisition activity, you need that kind of ability to quickly ramp up and spin up something, get that to its, you know, its stair stepping, get that to its kinda stasis, and then you can bring that to where it fits better. Maybe repatriation. It may be just a different flavor within the within the cloud. But that’s that’s really what I see and and how I have those conversations with clients here at Deft. What got you where you are, and what what made you kind of lift your head up and say, this may not be right.

We need to talk to somebody. And, you know, sometime it’s sometimes it is just being a digital therapist. You know, it’s it’s you know, talk to me about your problems, and and then we’ll figure out, you know, how you got here and how we can get you to the place that you wanna be.

Max: 40:29 I use that line so much. It is so funny that you just you just followed that. I I really I really explain, like, what I do. Like, you know, there’s, like, the drunk answer of, like, you’re what do you do? Right?

And it’s like I feel like a marriage therapist for, you know, you know, people buying technology and and companies trying to sell the technology to them. Right? And you kinda sit in the middle and, like, so funny.

Byron: 40:53 Yeah. They come they come to us with sometimes they have they they have what they think is buyer’s remorse with the cloud, and it’s no. This actually served you really well for what you were doing and how this this piece of your your business works. But it doesn’t have to be that one tool that you you throw at every single problem for you. And that that hybrid cloud solution is you know, that’s literally where we’ve been.

It it is where dev chooses to operate. It’s not easy, because every solution that we work with our clients is bespoke. You know, it is crafted for them for their specific problems. It’s not a you know, here’s here’s 3 different LEGO bricks, and you can really only build, you know, certain things with those with those bricks. It’s let’s see how all of these can be assembled to to solve your specific problem.

Max: 41:40 Let’s talk about FinOps for a little bit. So, you know, the Borg is dubbed as FinOps because everything either has to be as a service or something ops. Right? And and I think just you know, traditionally, in most companies look at cost containment as a finance function. Right?

Like, you know, ultimately, this goes up to the CFO, and the CFO is the one that says, you know, no. You can’t spend money or this is how much money we can spend and and tries to keep the company and and the business in line with what their expenses are. There’s almost 0 levers that finance can pull to influence cloud spend. I mean, you you can you can negotiate an EDP, which by the way, I hate EDPs. You can purchase savings plans.

You can buy our eyes. You know, there’s there’s there’s a few levers that you can do that become, like, exercises that you can have on a spreadsheet to try to do cost optimization. 98%, 99% of all cloud costs come back to an engineering decision of how an application was architected and then is run. How do you help a company go from, like, this idea of like, oh, we’re we’re trying to figure out how to use CloudWatch because we can’t understand our AWS invoices to, like you know, I wouldn’t even say, like, to, like, the holy grail of, like, they actually can can do cost accounting based on an application to, like, what’s what’s by the way, I have clients that can do that. And when you see that the first time, when you come across somebody that can say, this thing that we’re doing cost the company this much money, and we know exactly what that is to the entire infrastructure stack and how much revenue we’re driving from this thing, you’re like, it’s it’s it’s probably one of the most impressive things I’ve ever encountered in my life in terms of, like, what the organization did.

But anyways, that’s like holy grail Nirvana that probably is like like, you know, like, the 0.01% of businesses. What, how do you how do you take people on that journey? And and what does that really entail for for most companies that realize they need to get serious about this?

Byron: 43:34 It’s it’s the mapping process. You know? They they know what the bill is at the end of the month, and then it’s starting to, you know, dig through those layers. Alright. Well, let’s let’s do the cost analysis.

They usually have CloudWatch. Like, they’ve got all the tools in place, but they aren’t really digging in and have the expertise to understand the output they’re being given. You know, it’s it’s numbers on a sheet and maybe they can do a you know, they can boil it down to, hey. Well, this this group within the company has these products and I can associate, well, at least this little slice of it. But I I agree with you.

Having a company that can take their AWS or or cloud bill and do cost allocation and have the the the revenue generated behind that is is it’s a rarity. In fact, I I would love to see who you’re seeing because I probably have never seen that in my life, because it’s that it is that difficult. And and, really, you know, here’s the top level. Let’s figure out what what products are in there that that are driving that cost. And, you know, from there, it’s, you know, do any of these surprise you?

Let’s trend these over time and then keeping going one more level. Okay. Well, you have a bunch of, you know, these instances, and they’re doing what for you. And can you map these back to business processes, business objectives? That’s when you do uncover the well, that business objective was turned off, but no one ever turned off the servers or let us know that no one was doing x, y, or z anymore.

And, you know, those are the fun quick wins. But then it’s the the design phase of of coming back and say, alright. Now we have a ton of data. We understand how this all maps to your business. Let’s talk about what is priorities for you.

You know? And and we can rank those as as far as what is important to your business, and there’s some more levers we can pull. At the end of the day, our clients understand their business better than we do. And, you know, we strive to to be that partner with them, but at the end of the day, if that’s what they do, best, and they’re looking for a partner that can help them understand how their IT spend supports it. Because at the end of the day, they’re okay.

If if a dollar spent makes $2 on the other side, you can justify it. But it’s it’s sometimes it’s you’re breaking even. Sometimes it’s well, no. You’re actually spending quite a bit more money over here. You’ve got an EBS volume that someone forgot about.

You’ve got a developer that spun up, you know, the x 2 x large, 3 x large instances to run something like RabbitMQ, which doesn’t need that. You know? Those are the easy ones to kinda start picking apart, but that that mapping exercise and I and I see that exercise in in another way that we work with our clients, which is, you know, business continuity planning, where you’re really marrying that IT need to what the business needs. It’s a very similar operation. You know, can you map a business process to this expense and understand the value driven at the business minus the expense is hopefully, you’re getting a a positive result out of

Max: 46:30 that. A lot of people have a story or an experience of we made a misconfiguration. Whoops. We spun up a lot of instances or we did something, and we all of a sudden have, you know, a $60,000 bill from 2 days that we weren’t expecting to have. Right?

Like, that’s that’s a pretty consistent and common experience. Maybe it’s, you know, like, what scale it is? Like, who knows? And I kinda liken that to, like, you know, a sprinkler head in your front yard breaks, and, you know, you can kinda see water spraying. And and it and you become very aware of it very quickly, and then you go do something about it pretty fast.

Like, the the total amount of impact to you, you know, it’s it’s painful on a very short it’s a very acutely painful in a very short amount of time. But there’s all these other things that you kinda run into. It’s more like a pipe breaking underneath your house that you won’t see for months. And the amount of loss that that and damage that does to you over a long period of time is really, really substantial. So that company I was I was telling you about that actually it was impressive because they were able to make business planning decisions of we’ve we’re running this.

You know, we have the service and it’s generating this much revenue for us. And this is what our cost is to maintain the service and to run it. Explorations of, like, could we rearchitect? Could we rebuild? Could we adjust our cost basis for it?

And there’s one specific specific example where they decided to eliminate a service line, you know, from their from the business because it was just like, it’s not worth it for us to do this. It’s making and and and this is not like a small thing. It was 1,000,000 of dollars a year worth of revenue, but 1,000,000 of dollars a year worth of expense. And it wasn’t something that was really core to the business of what they wanted to be in the in the business of anyways. But they had that information to be able to do really, like, high level c suite executive planning and saying, you know, like like, what is the business driver?

And, you know, you kinda talked about this earlier. You know, IT is a pure IT is a procurement function and the cost of and time with the procurement function and the expense associated with IT. And I’ve said this for years. Like, if you’re in an IT team and and the company perceives IT as expensive, like, you need to figure out how to change this conversation with your business because IT doesn’t spend any money. Right?

Like like, there’s not a 1000 employees working for IT in a 1,000 person company. You know, there’s there’s probably only, like, a dozen employees in a 1,000 person company working for IT. Right? Maybe. Maybe.

Maybe. Maybe there’s only 6. Right? So there was, like, 6 people didn’t randomly buy, like, a 1000 laptops. You know?

And and but this this just if that’s the discussion you’re in, you’re, like, in the wrong discussion because it’s like, no. No. No. No. Like like, marketing has a 100 devices and and these applications.

And, you know, sales has has cell phones and and these services. And, like, okay. Fine. We we manage it as a centralized procurement, you know, function for the business because that’s our job, but it it it’s not our expense. And and I I look at I look at cloud in the very similar way of, like, you know, a lot of companies just kinda group this together and say, okay.

You know, the tech the the, you know, technology is spending all this money on cloud. And you’re like, no. No. No. Like, that’s not actually what’s happening here.

The CIO owns it or the CTO owns it. We’re just we’re just running what the business needs to run-in order to function. Right? Like No.

Byron: 49:50 It’s I I think IT teams, many cases, are are are on the defensive to justify, the costs or the spend rather than taking the kind of lean into it and and showing the the opportunities that it’s affording. You know, where you’re actually enabling revenue on it is a conversation that’s much easier to have. But also admitting there’s always opportunities to optimize. There’s always opportunities to say, well, is this a priority to the business? We’re trying to keep things running.

You know, we got a small staff. It’s 247. You know, there’s always the the horror stories about running IT. You know, it’s it’s not an if things are gonna break. It’s when things are gonna break.

You know, if you haven’t spent a a night in a data center trying to figure out why something won’t boot. It’s it’s humbling, to most of us that have that have experienced that multiple times, but it’s it’s showing that value and and goes back to that opportunity cost. You know? It’s there’s a cost of not doing something, and how does that impact the business? I think that is something that that a lot of IT teams struggle with.

It is really justifying the expense because it is directly correlated to revenue being generated by the company on maybe, you know, bring back some financial terms, back of house, front of house. The income coming in is enabled by many of the things that the technology is is bringing to the table. But I do feel like, at times, there’s a large disconnect between those two teams, and that mapping process will there’s light bulbs that go off. There’s commonality. There’s there’s an empathy that you now can have a CFO and a technical person in the room, and they’re speaking language that both of them understand.

Like I said earlier, you know, worst thing I can do is get on a call with one of deaf clients, and I dive too deep. I’m talking too technical, and it’s it’s the wrong audience. You can you can honestly see them unplug. You can see them disconnect on a call. They aren’t asking questions.

They aren’t looking at the screen. You could tell they’re off on Slack or something else. So that’s why it’s it’s important when we talk to our clients is to engage them as quickly as possible, and then let them take us on the journey, and and be kind of that digital therapist or or or guide.

Max: 51:55 I I laugh because I’m I’m guilty of this. I come from a place to think of also education, especially when you’re dealing with technical teams that then have to own the solution that gets deployed. Right? You know? So, you know, I’ve, over the years, end up defaulting to, like, we’re gonna talk about this and educate you on the differences so you can make an informed decision.

And, I can remember one of those meetings where I just I knew that it completely glazed out, and I finally I don’t know why. I just finally said to them it was something like, do you wanna know all this, or do you want me to just tell you the answer? You know? And and I think it was shocking to all of us. Right?

Because, like, you know, the the the people I was on this meeting with, you know, they kinda like like like looked at each other, and they kinda looked at each other. And then they looked back at me, and I was like, no. No. We just wanna know the answer. And I was like, okay.

Fine. You know, I I like in this also, though. I mean, the sense of, like, cost, you know, so, like, in cloud. Right? Like, this comes down to tagging.

Usually, you know, tagging is a a conversation that you have first. Can do you have do you have your resources tagged? And can you use those tags to create groupings and and do cost, you know, analysis against. Right? And that’s a really kinda crude description of this.

And I liken it to unit testing and software. You know, unless you started from scratch, unit testing, you don’t have any unit test coverage in your software. And companies get into these conversations that are, like, all or nothing. Like, oh, we in your software. And companies get into these conversations that are, like, all or nothing.

Like, oh, we can’t possibly have a unit testing for, you know, like, coverage because it’s gonna take us months to do all this unit testing. And and it’s like, no. No. No. No.

No. That’s not the objective. The objective here is you just have to decide as an organization that going forward, you’re gonna add unit test any piece of code that you touch. A year from now, you’re gonna naturally have a significant percentage of coverage of your code base in unit testing. Like and it just it just happens day to day.

Like, you know, you start at 0, and and after the course of a year of just fixing bugs and releasing features, you’re gonna be at whatever x percentage. You know? And and the hyperscalers do something similar when you look at if you’re using any sort of infrastructure as code, if you’re using a Terraform, if you’re using CloudFormation, if you’re using just an auto scaling group, you can just tell it to, like, tag new resources with a tag. And and, you know, and and by the way, here’s a really interesting kind of technique with this is at the end of 6 months, 8 months, 9 months, a year, whatever it is, you can find resources that haven’t churned over that have been static for that period of time. And it’s also an interesting exercise to say, okay.

What do we have this thing here? What is it doing? Why is it static? You know, do we really need to have this in a cloud platform if it hasn’t changed, you know, over the course of a year? And, like, just start.

You know? Like, don’t don’t talk about this to death. Like, just do something and start because you’ll never finish. Like, this isn’t like, like, there’s no in state with with cost op management. Right?

Like, it’s just this is a continuous process you have to run.

Byron: 54:30 Yeah. It’s it’s it’s, you know, process discipline is another thing we talk about here at. You know? You gotta put the process in. It’s going to be painful at first in many cases, but, you know, so starting, you know, training for a marathon.

Like, you’ve you gotta start somewhere. You’ve gotta adhere that discipline, and you will find that, yeah, 6 months on, you’ve built that muscle. This is easy now, and you’re able to focus on something else. But that that’s kind of the thing that I’ve seen just over the last 3 years with the pandemic, that operational best effort that I was talking about kinda set some teams back. You know?

All of a sudden, they were they were not in the same location anymore. They were working remote. And how do we work remote? And there was a lot of a lot of things flung at the wall. And now we see teams being able to take that moment to talk about it.

And then there’s also the the pressure of of optimization that’s going on. You know, everybody’s emerging from a a pretty wild 36 months of of trying to support, business operations and a new kind of, you know, hey. Everybody’s worried about where your data’s gonna go for the longest time. How do I secure my data? Well, all of a sudden, my people are in places I never imagined they were gonna be.

So all those resources were were a challenge, but it is. There’s there’s simple things you can do to start building that muscle right now. Every incremental thing that you can you can do will add up, and it’s pretty shocking. You know? If you can improve one thing each day, you you just build upon that.

And but that starting piece is hard because I think a lot of times, companies are coming to talk to to DEFT about how do we start? You know? How do how do we you know, I need a partner. You know? You jump, I jump.

Like, let’s let’s tackle this together because it’s just that that confidence that they are embarking on the right direction. And in some ways, it’s it’s DEFT sitting with them, bringing our expertise and our experience to kinda point them more or less in the direction that they wanna go, and let them step off with confidence. And then also being there when that 5th step isn’t exactly in the right direction. Let’s pivot again. We’ve learned something now.

Let’s let’s yeah. You’re in AWS. Let’s let’s bring you somewhere else. Let’s figure out what this workload needs, to to operate.

Max: 56:36 Yeah. I mean, the therapy is a good is a good example. I also kind of, you know, use, like, car buying. I mean, you guys get to buy a car every day. Actually, you’re buying dozens of cars every day.

Right? You know? And to use this analogy. And, you know, your customer is buying a car once every 3 to 5 years. And the people that bought the last car aren’t buying the next car.

They’re usually you know, have churned off. And this is probably the easiest, like, leveraged creating thing that a company can do is accepting that even though they have people responsible for driving the car in house, like, it’s okay to have an expert help you buy the car that does that professionally for a living every single day, dozens of times a day over. And it’s not competitive, and it’s not a bad thing, and it’s not it’s not a negative. You know? It’s like, who cares that, you know, your VP of blank didn’t run the negotiation or or x, y, and z in order to buy the car.

Like, do you do you do you is that what, like, you value? Or do you really just value did you get the best possible outcome for your business? And, you know, you you said it earlier. You know, there’s endless amounts of tools in these platforms. And even when you start talking about cost management, AWS natively gives you a bunch of tools.

There’s lots of third party tools. But who’s gonna run the tools? Who owns the tool? Right? Like, so you you know, like, you know, a 247 shift cycle is 7 bodies.

It’s 7 people to 1 run 1 247 shift. You know? And then how many how many people deep at every given shift do you need at well, you need more than one body. Right? You need probably 2 or 3 people.

Right? So, like, you just multiply this up. And is this something that you wanna invest in? And is this generating revenue for the business? The answer is usually no.

You know? Your point earlier also I come back to was, like, most companies are not in the business of IT. Like like, IT and technology needs to be a a business enabler for those businesses, but go out and use the experts that do this stuff for a living, who created runbooks, who know how to document process, who can deal with hiring, training, attrition, churn, you know, like, all these sorts of things that can you know, you just can’t do at scale.

Byron: 58:45 Yeah. That is exactly where where you know, when I was a former, former client of Devs. That’s exactly why I came to them. You know? I needed to get out of a data center that we were running in in our building.

It’s very expensive per square foot. You know, you’ve talked to a facilities manager running your own data center in your own building is

Max: 59:03 Bad idea. Don’t do it. If you have your own data center, do not do not continue. Go to a data center. This is for a living.

Byron: 59:09 Yeah. It it’s the the the amount of redundancy resiliency, the the our data center operations team is 247, incredibly talented group. This is what they have a passion for and what they love to do. Is your is the company’s team better off paying someone just to change that hard drive when it fails, or do you need a team that, you know, is gonna handle that and just tell you, hey. That hard drive failed.

We replaced it already. Like, they’re ahead of the game, and that’s what we bring to the table. So it’s more of a you know, I we talk a lot about races. If you could you you can get away from being responsible and accountable for a large portion of that infrastructure management, whether it’s cloud or or on prem or hybrid. What does that allow your teams to focus on and unlock?

I think what it does is it allows those teams to drive more business value and show more business value and not have to justify why we were changing hard drives, you know, a couple times a week or jockeying tapes or staying up all night with teams that are doing a firmware upgrade on a, you know, Cisco UCS platform, which I’m I’m labeling literally everything that I used to do, which is why I came to Daft years ago and and and went went into that managed services because it’s incredibly draining on a team, especially small teams. When you talk about 247, most companies don’t have 247. It’s they’re there during the day. You get a call at night or you have something which disrupts business operations in those teams for the next day. Whereas we’re staffed up, we’re ready to handle those kind of challenges with multiple 247 teams to to do those things for our clients.

Max: 01:00:37 And this is of course, I do not make this example as a negative with theft in any way, shape, or form. I use it to illustrate the point of, like, know what you’re good at. Theft has been in business for over 2 was 23 years, 2 decades plus. A good percentage of your data center infrastructure is leased from REITs that are in the business of building and managing data centers at $1,000,000,000 scales. Right?

So your value add isn’t that you’re investing 1,000,000,000 of dollars in building a a concrete wall tilt up powered shell data center. Right? Like, a REIT can go capitalize that and manage it and have you know, go through the entitlements and everything else that goes along with it, and you can operate the equipment on the inside of it better than they can. And this is I don’t want you to say, like, build versus buy because that’s way too pedestrian of an example here. But, like, there are so many different places where you can drive efficiencies and leverage at scale that benefit you.

Know, like, running a data center in your own premise, like, we helped I mean, I had a hospital do this that wanted to do it and, like, would not the c I mean, really, the CTO, the CIO wanted on on the resume that had a data center that they built a data center. Right? So they had this, like, 400 kilowatt data center that they built. They spent so much money. And there was a carrier neutral data center 2 miles away.

Like like, there was fiber running in front of their building. You know? It would have cost them it it was it was like their the cost that they spent building the data center, not even operating it, just building it, would have paid for them to colocate for, like, 7 years. And what value did they get out of having the data center? And there’s nothing.

They got no value out of it. They they they burned up some spaces, put up some parking spots, you know, to put a generator in it. Then they had to get permits from AQMD in order to run and test once a month, which they didn’t do because the permits were expensive. So they only tested it once per year, and then the generator failed, yada yada yada.

Byron: 01:02:13 Yeah. But though those are the costs. Those that’s the life cycle costs of doing something like that. Whereas, you know, Deaf partners with, yeah, some of the the the top data center providers in in the world, but we bring that management of the floor space to our clients and provide them with, you know, I won’t say infinite, but, you know, every solution that we deploy on our on our data center floors or in AWS or a hybrid solution is very customized for what that client needs. It’s it’s not a, you know, racks and racks and racks of servers that are all of one flavor, and, you know, we gotta figure out how how we can shoehorn a client’s workload into exactly what we have on hand.

Max: 01:02:52 And and for clarity here, DEV’s portfolio, when we talk about services, you help people manage a hyperscale infrastructure, you know, in an AWS. And you go from there all the way down to you know, somebody wanted to to license colocation space and do their own server management, they could do that with you. And so, really, in between those two extremes, you have, you know, managed managed colocation. You have, you know, bare metal dedicated servers. You’ve got your own hybrid cloud platform.

Or you do private cloud platforms. You have a, you know, a public cloud platform. You know, there’s there’s lots of different permutations of how to do this. And and I wanna just kind of, like, point this out here because you say, like, everything becomes bespoke. I mean, a customer could choose any one of 4 or 5 different places to run a workload, and you can support them from point a to point b and then move them from b back to a and then, you know, over to e or whatever it actually is.

Hopefully, you’re not doing that with somebody every, you know, 6 to 9 months, and and they actually stay in place for a while.

Byron: 01:03:57 Because Not the same same company, but, you know, we we find companies that are on that journey, you know, at any point. One of the common threads that I’m seeing now is because of remote work. You now have a data center that’s in a building and where there used to be 250 people showing up every day, there’s 5. The data center that’s sitting there is now there’s a very technical term, but a boat anchor for them to be able to maybe downsize leases or get out of leases, and they’re struggling to figure out what they need to

Max: 01:04:26 do. And and Really interesting point. Yeah.

Byron: 01:04:28 We we can we can absolutely, you know, find a quick solution that will be, you know, move them into the data center, but we know that that’s not the end of it. That that is part that is the first step of that journey where I was talking, like, sit you know, guide them and partner with a client to take that first step to unlock a potential. Maybe it’s saving a ton of money by, you know, downsizing their their office space. But at that point, then we’re we’re looking to optimize. Hey.

You know, you’ve you’ve got a server refresh coming. What if you didn’t have that CapEx spend? What if we get you into an OpEx model? We have private clouds. We have Detties.

We have multi tenant VMware. Oh, by the way, we can also if we find a workload that that would work better in AWS or Azure or GCP, we can work with those. And I mentioned earlier, that’s not a one way street. You know, we have a client that moved into, you know, using quite a bit of of, GPU resources out there. Company called Pixart, not too small, using a ton of GPU resources with 1 of the hyperscalers.

They just moved quite a large, purchase of Lambda servers, you know, the big GPU servers, into our data center, you know, because of the cost analysis. They were able to do it cheaper and more cost effectively that way. If you ever seen these servers, they are they’re insane. I it is the loudest and hottest thing I’ve ever seen in a data center. If you walk by outside, you hear them, and it’s a little it’s a little disconcerting.

Max: 01:05:47 I’m gonna harp on this part. These are not all or nothing decisions. Nothing that you’ve talked about is an all or nothing decision. Purchasing a a dedicated GPU farm that you put into a data center is not an all or nothing decision. You haven’t decided that the only place you’re gonna run this workload is on physical equipment that you put into a data center, decided that for x percentage of this workload, we have this equipment that’s gonna have this economic model for us, which, by the way, is phenomenally better than the cloud vendor is.

And if we need more than that and you actually need the elasticity, we can go and we can go into any less you know, into a cloud and we can burst capacity. And, by the way, you know, the most common use case I see with multi cloud today, Bigtable and BigQuery. And the driving factor for Bigtable and BigQuery, fixed or flat rate pricing based on a dedicated consumption. So this is as anti cloud as it gets. But when you think about it from a, we had a workload in Redshift, and it was very expensive, and it was unpredictable, and it was variable, and all these different things.

And so instead, we went and decided to buy 2,000 slots of capacity with with big, you know, with BigQuery and and shift everything over to BigTable. And we now know this thing is only gonna cost us x tens of 1,000 of dollars a month. That is an that is the the exact point of should you even be running that in a cloud platform in the first place, or is there a different workload that makes more sense for you? Because, I I mean, there are lots of workloads that don’t need to be instantaneous response. Right?

Like, you can say, we can have a job queue. And if it takes somebody to get instead of having it instantly, they get it, like, 45 seconds later. Like, it’s not the end of the world. I mean, look at look at basically any SaaS platform. You ask it to do something and you get a response that says, we’re gonna email you when this is ready.

That’s just optimization of of consumption of resources. I mean, that’s smart that’s smart business.

Byron: 01:07:37 Yeah. We even this client that we we’ve, you know, with PicsArt where we we brought those in, they still run workloads. It’s not like they turned one off and went, you know, it’s a 100% this or a 100 that. It’s never an all or nothing proposition. You may have some some workloads you wanna prototype quickly, proof of concepts.

You know? And it is valuable to run that quickly, get your data so you can make a data to data driven decision. It’s something we talk about here at Daft quite a bit to to move to the right position, even within our products, you know, our multitenant cloud, that we have, versus a private cloud. We know where the inflection points are. Go talk to our clients.

You know? Hey. You’re you’re in a private cloud, but you actually you know, you might be better served to move some of these workloads to our multi tenant. So even within DAFT, it’s it’s never a one service will always be the right solution for it. You know, we’ve got inflection points when you should grow too.

You know, it’s you hit x number of storage. The cost numbers usually look better when you’re, you know, maybe getting a dedicated SAN. But it it it affords us an opportunity to talk to our clients, and that’s really the the core of what we do is is making sure we understand what our clients are doing. We’re talking to them and partnering with them, not just waiting for them to call us up and say, hey. I need more of x.

I need less of y. That’s never a conversation that I like to have. You know, it’s what’s changing? What keeps you up at night? What do you what what are the new OKRs or business objectives that you’re getting asked to implement or affect?

You know? And and those kind of conversations for me, here at is a lot of fun to have those conversations. You know? But getting the whittling that down from the art of the possible to, you know, what we’re going to do, there’s a lot I’m I’m hand waving a lot, in that statement. You know?

It’s it’s somebody that can can at least come in with with the muscle built to help guide, and it’s very consultative. It’s very consultative, at times, and it can take a while to get to some of the solutions, but it’s never a one size fits all. That’s that’s the thing that I really love about Daft is it’s not a, oh, you need to buy 5 servers, and here’s your quote and there you go. Happy to do that for you, but I got a lot of questions. I wanna understand what we’re trying to do, how this impacts your business, and what we can do to make, you know, an IT team’s life easier and help them show the value that they bring into their company.

Max: 01:09:55 Yeah. So, I would normally I would say, like, anything that you wanna end on. I mean, that’s probably already a pretty good ending. So other than the obvious, if you’re having problems with any of these things, you should talk to Byron.

Byron: 01:10:09 Absolutely.

Max: 01:10:11 What else, you know, what else should we really wrap on? Like, what’s the what’s the 32nd thought to to cut this off?

Byron: 01:10:20 I think one of the things that I I wanna impress on folks is is it is not an all or nothing. You know? The the resources and tooling you have from on prem to hyperscale or cloud is a toolbox that you, you know, you get to pick and choose the pieces that you need that’s best for your business. And here at Def, like I said, we we were we were hybrid before hybrid was cool, because we knew it wasn’t going to be all or nothing. And and when you feel like you’re locked in or you feel like there’s a lot of inertia behind something and it’s hard to change, I’d I’m happy to sit down with folks and talk to them about that and and have that conversation.

Do it pretty much 3, 4 times a day with clients.

Max: 01:10:59 Amazing. Awesome. Thank you very much, Brian.

Byron: 01:11:01 No problem. Thank you.