Introducing Edgevana - Mark Thiele

Machine-generated from the episode audio and not hand-corrected, so names and technical terms may be imperfect. The audio is authoritative.

All right, you're listening to episode insert, whatever number you would like here, because it doesn't matter. You can go back into the archives and see. We already had a conversation with Mark Till, one of the most well-known experts in the data center space. So we're in for a treat, not just because we have Mark Till on again. He's now coming back to us as a CEO and founder of Edgevana. And we're going to find out exactly what Edgevana is. We had Mark on a CTO advisor edge computing panel.

You guys love his insight from there. But we're going to go to kind of his sweet spot, which is I think Mark data centers. I think the last time we talked, the original idea was data centers. But what are you doing, man? Yeah. So thanks, Keith. And again, thanks for having me on the show. I mean, I know that there's plenty of good content and folks out there. So I appreciate you. You bringing me on. So Edgevana was founded under the notion that, I mean, really kind of several underlying themes.

But the but the biggest underlying theme was that there's going to be incredible demand for data center capacity of all types. And that data center capacity could be anything from something that looks like an ATM in the closet of a warehouse in any town USA or any town South America to something that's got, you know, 500 KW or five megawatts or 50 megawatts or 500 megawatts in the core with somebody like Google or Microsoft. And basically everything in between. And, you know, you and I have talked about this.

Certainly I've written a lot of content about it. And it's not like I'm the only one that suggests this, but the the volume and the demand is likely to outstrip everything that is available today. And so considering that I have a little bit of a sustainability bent to me, plus I well, I'm known as a data center guy. I'm also known as a cloud guy and I'm trying to find the right way with Edge to marry those two as much as possible.

And to me, the marriage of those two is a number of things. One is is best fit. Right. So instead of making an assumption that if you're doing X, you don't have any choice but to use Amazon or Microsoft or Equinix or DRT or or HP or IBM or something or Intel versus AMD versus Nvidia. Rather, that best fit means that you can use the appropriate and best options to solve for your business problem. And, you know, the epitome of historical pragmatic decisions we've had to make are are leaving out members of the community because the network expense would be too high or brute forcing a technical problem with X86 when GPUs or FPGAs or some combination of the above might have done a much better job.

And so I'm trying to help solve that and make a significant amount of infrastructure that is largely going under the radar, invisible at the edge and just globally distributed data center capacity available in a much more agile fashion, much more consumable fashion to the buyer. And in a simple nutshell, historically, if someone wants to get data center capacity, whether it's for Edge or for Colo or, you know, for one or for 10 locations or more, the due diligence effort, if you do it yourself, involves months of work and weeks or more of contracting.

And so even just getting one new site can be a three to six month ordeal, let alone trying to be in 10 different countries or, you know, 100 sites in 50 different countries. The associated with doing that is left only to the largest of players. And even then, it's a huge burden on their cost of doing business. So I'm trying to help solve for that and lower the barrier to entry for people to consume what's available already, because the most efficient data center ever built is the one that you don't have to build.

So how do we best use what's already there? How do we lower the barrier to entry for the consumer to get access to the market and make it more cost effective and more agile? And, you know, I realize that's a little bit long winded, but in effect, that's what Edgevana is starting out to do. So, Mark, one, I absolutely see the challenge. I'm coming from, you know, you know, my background. I'm primarily in Fortune 500 companies. And while people think that Fortune 500s have unlimited budgets, they typically do.

But that doesn't solve every problem, like having the budget to be able to spend money with an Equinix, a Quest or QTC doesn't really matter. Insert tier four three data center here. The problem is, is that they're not everywhere that I need them to be. So if I need colo space in the middle of South Dakota or in Malaysia or wherever, as a architect sitting in my cube or office and an Enterprise 500, I might as well be a small fortune, a small SMB, because the size of the problem is still the same.

There's this thing. There's this data that I can't surface. We had a vendor on a couple of weeks ago that talked about helping with this problem. But I think your angle to this problem is completely different where they were mainly consultants helping you surface up and bring the relationship. Talk to us a little bit more about the end user experience. When I'm the architect and I'm coming to Edge Vinyl, I'm the CTO, CEO, CIO of some company that needs capacity in a that that has a certain size and shape to it.

What's the experience? Yeah. So there's logically there's two different experiences that lead to the same thing, maybe without the same level of automation in the first experience. And so, you know, part of our discussion today is about the fact that we'll be launching in a soft launch our Web site with much greater visibility into what it is we're trying to do this coming Wednesday at 9 Pacific. And we'll be pushing out some news associated with that, etc. But the difference between what we're offering on Wednesday and what we'll be offering when we launch the platform in Q1 of 2021 is that there's a lot more in the first version.

It still looks like a duck on the lake, but there are like six sets of legs running at full speed underneath in order to try to make what will be automated appear more automated to the consumer of our services. So behind the scenes, we'll be doing things to help pull contracts together to make multiple contracts with one with, you know, from 10 suppliers seem like one contract in overhead and time consumption for the buyer. The ability to pull together a combination of services to solve for, you know, a full stack requirement.

And I don't mean just full stack hardware wise, but full stack from total demand of service enablement for an edge deployment. As an example, you know, you might need data centers in a bunch of locations. OK, we'll help you do that. You might need network services brokered for those locations. OK, we can help you do that, too. You might decide you want SD-WAN capabilities between those locations. We'll help you do that. You might decide you need smart hands because you're going to deploy your own hardware and you want somebody to take care of it.

You might decide you need some help in hardware acquisition. You might decide you need some help in hardware automation. You might decide you need some help in capacity planning and AI ops for those environments. You might want to just hand it all over and say, Mark, go build it to the spec and deliver it to any of the locations that meet my SLA and we'll do that. And so that's longer term. A lot of what I just described from the platform will be supported by automated search and selection choices in the platform.

But initially, it'll be a little bit simpler in the sense that customers can come in, go to the site, get an idea for what it is we're doing. Leave us a note and we'll get back to them almost immediately and help them get started on their journey. And again, it could be as simple as finding one data center in an unusual place. If we don't have a data center there with a current supplier, we'll get one. It's been relatively easy for us to add new locations and capacity to our footprint.

It's been really amazing the kind of positive feedback we've gotten from the data center community as a whole and from the customer community. So I'm really excited to make this available to folks and to make this part of the infrastructure world a little bit closer to being agile, like you would expect from doing something highly virtualized or buying from a cloud supplier. Yeah, I love the initial concept because going from, you know, if you're a old school network engineer or data center engineer and you've ever had to provision redundant circuits coming into a facility and set up routing and all of that and then go to software to find networking in the cloud.

And you're doing it with just a couple of clicks of buttons. It's like an amazing experience. But the gap between that and still doing this low level stuff that's not in the public cloud is so huge. So what I'm hearing you say is that you're looking to close that gap a bit and say that, hey, you know what, some of this stuff should be abstracted. I shouldn't be a architect spending my time calling eight different providers in 12 different regions to get a set of services.

I should be able to go to a site and provision. I know exactly what it is that I want. I just need to find the network providers to do it. And I don't want to manage the individual contracts. That's just something that should be real rolled up. Somebody should have created this already. Exactly right. In fact, I have that question asked of me on a regular basis by anyone from a potential investor to some of our first partners is, God, I can't believe somebody is not doing this already.

And so it is really kind of a surprise, to say the least, that no one has done this yet. But, you know, to your point, why would we make any assumptions if we're telling, you know, the Keith's and Mark's, you know, both of us have had our years of pulling cable and cutting fingers on on on sharp server cases and shit like that. Pardon my French. But, you know, now there's no question that that when we would make recommendations to people to find ways to abstract the kind of work that we used to do in setting up a data center.

And so when I look at global deployments, if you are going to own the deployment, what's important is is is building the deployment and deploying it. What's important or is ensuring that the service that that deployment is meant to enable functions to minimum requirements and and and better on the schedule and on the budget that you're worried about. To me, the results are what are what matter. Right. And so I don't see why what we're trying to do should be considered any different from, you know, providing some automation to somebody in a data center or telling somebody to take that particular workload and putting it in the cloud.

You know, if I'm going to put something in the cloud, I don't call Amazon or Microsoft and say, OK, tell me exactly how you rack that server. Tell me when you've plugged the server in. Tell me what kind of cable you use. Are you are you using green or blue for Ethernet versus command line, whatever. You know, you're not you're not asking those kinds of questions. Those things have been abstracted. And so I'm hoping to make much more of this very physical world abstracted and more accessible to folks as they look to build out their their solutions, whatever they may be, whether it's a single site, 10 sites or a thousand sites around the world.

So talk to us about these partnerships and the difficulty, because I think as I think through what it takes for me to contract with just one partner, I it took me maybe five weeks to sign the contract for my data center. I only have three racks and office space, and that took me five weeks to get done. And it was one with one of the larger providers. Right. What's the level of effort to contract with each one of your providers and what what are you cutting through for your customers?

Yeah, the hard part for me to actually try to promise Keith is that I'll be able to take, you know, four weeks out of what might normally take five weeks with one vendor. That's the hard part for me to promise, although it's definitely a big part of what we're trying to solve for. But as important, maybe more important, depending on what kind of need you have, is that instead of making five purchases, each have additive effect contract negotiation periods, meaning, you know, if you contract with one and you finish it, you start another one.

You got five weeks, followed by five weeks, followed by five weeks, or even if they overlap a little bit, you're consistently adding another week, another two weeks overall to the delivery of your, you know, one, three, five, 10 locations. Instead, try to make that five week period shorter and make it not much more than five weeks or whatever the shorter period we can define the same, whether you're contracting with one supplier or 10. Right. That's really the goal. And then I have the added benefit that the SLAs look awfully similar, if not the same, very similar for the tier of service that I'm getting.

So if I want a tier one, two, three, four data center, you guys know what my requirements are as you go out and work with your partners. You essentially become an intelligent purchasing department. So in large enterprises, we have purchasing groups that do this work for us, but they're not data center experts. They normally come back to me with, you know, a dozen questions that I could have just easily responded to the provider with if the provider sent the questions directly to me.

But that's just not the way the large enterprise works. So who's your, ultimately, who do you envision being your customer inside of the large enterprise? Is it purchasing or is it the technical team? Yeah, I think it will depend. I think in many cases, especially for first time out, it's likely to be the technical team. But over time, depending on how much automation we can supply and how much proof in the pudding we can support over delivery of a given period of time, we might get to the point where an IT team says, yeah, just go.

This is what we need. Go through Edgvana and requisition it. Right. And that's a possibility. But I think it's probably at least a few years away because, you know, as you well know, there are so many vectors for options within the IT world, even within just the data center, let alone within, you know, a specific hardware stack or, you know, time to deployment requirements, network bandwidth requirements, latency requirements, et cetera, et cetera. That taking all of those things into account, along with the nuances of, you know, have I bought networking from two different providers but found out that the networking is actually one provider because the second provider is using dark fiber from the first provider?

You know, things like that. Right. Those are things that are really hard for the average purchasing department to solve for. But depending on how successful we are in helping customers with, you know, defining their requirements, the sooner and the closer we'll get to helping to to make, you know, even complex purchases, something that you could hand off to an informed purchasing group. But I think initially the majority of our purchases will probably come from somebody that's, you know, head of engineering or are ahead of of data center group or infrastructure teams at companies involved.

So as you attempt to do this at the aggregate, you're obviously getting insights that these personas typically don't get. And I would love to kind of end off on that note. What has been one of the more interesting kind of aha moments for you? You've been doing this for like 30 years. Yeah. But even in this and your attempt to aggregate to kind of look at this big industry problem and solve it, what has been one of the more surprising or not?

It's not necessarily a surprising thing. It's kind of like, wow, I just did not have an appreciation for the level of complexity of this part of the challenge. Yeah, I think I think there's two things. One of them might be closer to the to the type and style of answer you're looking for. But I'll start with the other one that might not be so close. But it really is that so much of the data center community from an offering standpoint is is becoming I've I've found I sort of suspected.

But the surprising part is that I feel like I've proved it over the course of the last four months as we've really been aggressively building out what we're doing and approaching suppliers, et cetera, et cetera. And that's that, you know, the individual data center, even with five or six sites with a recognized and and and successful brand name are becoming more and more invisible to the buyer. Right. Even if they look for a replacement to the data center, they're looking for something that offers capabilities and services that would be unable to be provided within their own data center, whether that's managed service providers, whether that's cloud connection services, whether it's additional bandwidth and and and network service providers, you name it.

They're looking for those add on features that they can't do in their own data center. And then on top of that, you have the fact that companies, as they begin the process, I shouldn't say the fact the the the trend, at least as companies look to platform themselves, turn themselves into technology companies through business transformation. They're looking for more global partners. And so even if they feel like they could go to their local data center and buy from them today, they would be nervous to buy from the local data center provider because they don't know if six months from now or a year from now, they might need 10 locations in 10 different countries.

And if they do, they don't necessarily want to have a bunch of different contracts. And so can they get support and benefit from someone who can, you know, allow them to get into multiple markets without the pain that might be involved in doing due diligence on 10 different suppliers in 10 different markets? So that that effort, combined with the fact that more and more purchases are beginning to look like edge purchases or actually are edge purchases, both from a supplier to consumer standpoint and from a direct enterprise to consumer standpoint, has been remarkable to me to get that proof in the pudding from the evidence gleaned from speaking with so many customer potential customers and and so many suppliers to our platform.

And I guess the the other one is really the the complexity we've thrown in the process for. And you mentioned it already, the contracting piece. Right. I've been I even worked in the COLO industry for five years, let alone, you know, doing my own data centers and things like that. But I was I was amazed back then at the consumption of time involved in getting a contract signed. And but I hadn't looked at, you know, 100 different companies contracts and tried to compare them.

I just had an inkling that they were probably all mostly pretty close. And sure enough, I like to be conservative in numbers. So I went around and I talked to COLO operators and I tell them, yeah, it seems like, you know, using the 80-20 rule, probably 80 percent of the language in contracts are likely the same between suppliers. And every single one of the operators that I talked to and use that number, they say, oh, Mark, it's more like 90-95 percent. So we have this enormous amount of complexity associated with getting something done with people arguing over points that only lawyers give a shit about.

And the reality is, is that 90 percent plus of the contract is the same between suppliers. So what can we do to to help bring that uniformity to the base of the contract and abstract as much as possible the portions of the contract that are unique to the supplier or unique to the demand from the buyer? So, Mark, I really appreciate you taking time out of your busy schedule as you get ready to launch a soft launch in a couple of days. I wish you and the Edgevana team the best of fortune for those wanting to reach out to Edgevana.

What's the Web site that's going to be there in the next two days? Yeah, it's pretty straightforward. com. And certainly, if you'd like to find some material that will help you sleep at night, you can follow me on LinkedIn under Mark Thiele. And I'm fairly obnoxious and active on Twitter as well under mthiele10. And I love the interactions. For those of you listening, Keith and I, I think, share a lot in common in the sense that we like to challenge our assumptions and put our effectively open our kimonos online and get feedback and pushback from folks.

And so please interact, add comments to things I put on LinkedIn, challenge my assumptions on Twitter. That's the way I learn. Yeah, and definitely tag me in these conversations. I would love to be a fly on the wall or even get involved in the conversation to whether you guys agree with what Mark is doing or kind of like, you know what, this is a crazy idea. Everything is going. Kubernetes and cloud now magically fix it. That's one way to get me and Mark teed up for a really great dynamic conversation until the next CTO advisor.

You can follow us on the Web, the CTO advisor dot com. As my good friend Mark May likes to say, you can go ahead and give this podcast link to your mom, your aunt, your grandma. They will love to be put to sleep by the smooth sounds of Mark Thiele and Keith Townsend. Talk to you next CTO dose.