- Using AI at Scale with Adam Weiner of Granite - February 27, 2024
- Customer Experience Optimization with Bridgepointe’s Scott Prater - February 5, 2024
- The Evolution of IT Cost Management with Eric Russo of Bridgepointe - January 22, 2024
10 Critical Considerations for Your Next Data Center Project
In this guide, we’ll share the top 10 considerations to drive performance and cost savings with your next data center project.
On this episode of The Bridge, I’m joined by Bridgepointe’s own Mel Melara, Global SVP of Data Center & Cloud. We’re talking about securing data center resources and more.
Mel joined Bridgepointe in 2019 and brought over 30 years of experience in network and data center infrastructure consulting to the firm. As a sales leader at Masergy, he was part of the executive team that tripled the company’s revenues in five years. Mel also held an executive role at Switch Data Centers and is a subject matter expert on data center trends.
During our conversation, we talked about everything related to data center resources — what happens when hyperscalers buy up three years of inventory they’re now digesting, the challenges of sourcing data center space, and considerations when deciding where to go.
Topics covered in this episode:
- Mel’s responsibility for data center operators in the portfolio.
- Why choose Bridgepointe for data center space?
- Why there’s a heightened global demand for data centers.
- The influence of hyperscalers (AWS, Azure, Google, Oracle) on space demand.
- The enterprise trend of moving from on-premises to colocation due to remote work.
- Bridgepointe’s role in helping clients find suitable data center spaces.
- Covid’s impact on data center demand and overall acceleration of data center demand from various sources.
- The importance of data location for compliance, such as GDPR in Europe.
- ESG (Environmental, Social, Governance) initiatives and clean energy
- Bridgepointe’s role in helping clients find data centers with renewable energy
- Explanation of how carbon credits work in data center selection and carbon credits as a means of satisfying renewable energy requirements.
- The importance of the hybrid model in data centers.
- Bare metal as infrastructure as a service without the operating layer.
- 10 strategic considerations for evaluating opportunities.
- The factors affecting facility efficiency and gear longevity.
ABOUT MEL MELARA
Mel joined Bridgepointe in 2019 and brings over 30 years of experience in network and data center infrastructure consulting to the firm. As a sales leader at Masergy, he was part of the executive team that tripled the company’s revenues in a five-year period. Mel also held an executive role at Switch Data Centers and is a subject matter expert on data center trends.
Hi, and welcome to a special edition of The Bridge. I have with me today Mel Melara, who is one of our own. I’m thrilled to have you on the show today, Mel, how are you?
I’m good, Scott. How are you?
Fantastic. I’m glad you’re here. For those of you who don’t know Mel, I don’t know who doesn’t know Mel, but for those of you who don’t know Mel, Mel is our VP of Sales. More importantly though, he’s the head of our global data center practice. And that’s really the topic for today that I wanted to get into. So, Mel, let’s just start there. What does the global head of our data center practice mean to you and your role at Bridgepointe?
Yeah, so basically what it means is, we launched this practice like five years ago. Even though we’ve been doing data center projects for 20 years since the company started, we wanted to put together a process to take a client through, to help him procure data center space. The global piece of it really means I’m responsible for all of the data center operators that are in the portfolio. And so what I try to do, is I try to stay ahead of that curve. So just to give you an example. I’m recently signing a new data center provider that’s in Central America that gets us into Guatemala, Costa Rica, Honduras, Panama and all the other regional countries down there that we didn’t have a presence in. So I’m trying to stay ahead of that. And so today we have 55 global operators in the portfolio.
That’s fantastic. So I think that’s a great kind of next question, which is where people are sitting here going, why Bridgepointe for data center space? If I’m looking for it I could just fire up the internet and do some searches and find some Honduran data center space, pun intended. Obviously what makes what we do, specifically what you do, kind of special and how does that benefit customers?
Yeah, so that’s a great question. First off, the global demand for data centers is at an all time high. In the last 20-25 years, it’s never been higher. And I think there’s a little bit of a perfect storm brewing out there. As it relates to data center demands, you have the hyperscalers, which are AWSS, Azure, Google, Oracle, those guys out there buying up a lot of the space because their cost increased by 60%, their build costs to build the data center. So they immediately went and bought up the inventory, and that was over the last 18 months or so, and they ended up buying like three years of inventory. So now they took a ton of the capacity and locked it down and now they’re digesting that capacity. And on the heels of that, we’ve got this new AI-GPU boom happening where these AI-GPU companies are coming in needing huge chunks of space, 4, 5, 10 megawatts transactions. So those are very, very large spaces. And then you have the enterprises that are now working on trends like on-prem to colo because a lot of them are closing their operations or their facilities because people are working from home. And so that’s also feeding the demand. And all three of those are intersecting and colliding and there’s very, very low capacity out there. And the reason that we are so relevant and important in that process is that we know where the inventories are, what suppliers offer what services, and we tie that back to the client as we do a deep discovery, understanding their applications, their hybrid nature of cloud versus colo, and then we help them find a data center that best meets their desired future state when the thing is installed and running in production.
Got it. So reading that back, I mean, number one reason, value I should say, that you and your team are bringing to customers is availability. I mean, people are scarcity of assets, I guess is the best way to think about it, right? The big boys are buying space. We’ve got, as you said, 55 suppliers in a portfolio. You’re able to, in that respect, operate as a realtor. You’re finding the right spot, the right fit for that kind of thing, just with information that a customer would not be able to gain on their own. You did mention one thing there that was interesting about you know, the enterprise customer and their changes. I think people are buying the hyperscalers, I think obviously AI’s in the news, but let’s go in the middle of that and just talk about COVID for a minute. Is that what’s driving a lot of the kind of pent up demand that we’re now starting to see? The almost reinvigorated demand in data center space is companies kind of locking in on their hybrid work scenario and maybe skinnying their work campuses that may have included a data center?
Yeah, absolutely. And that trend is continuing to accelerate and COVID launched it but then you also have all the other stuff happening, right? Like people are home, so people are gaming, more people are watching Netflix, and all that stuff points to data that all ends up in a data center. So all of those trends are up and just the general demand for data center capacity is up even by the enterprise. And also the social media companies that feed the folks that are at home wanting more to use their tablets and their remote devices as well.
The Evolution of Data Center Infrastructure and Hybrid Solutions
You mentioned, and I’m bouncing around a little bit here, so forgive me, but my mind is spinning. I got you on here, so I’m gonna leverage the time that I got with you. You know sort of data location and data residency is important in areas like the EU. Is that beginning to drive? I mean, you mentioned adding Central and South America. Some of that is getting closer to your customer for streaming content and things of that nature. Some of it is regulation and is forcing data residency inside of countries. Are we seeing that as well? Is that starting to be a driver?
Absolutely. Right. DDPR in Europe, we’ve got compliance in North America, depending on the industry. Obviously financial services, healthcare, highly regulated in terms of compliance. So we definitely bring that into play. We have some government clients that we work with that have government compliance requirements you know, some that I can’t talk about today, but certainly there are important ones out there. So they are definitely driving where somebody wants to be and what type of compliant data center they wanna be in completely.
So you’re mentioning compliancy, so I’m feeling a natural bridge. Let me hit you with the next one. And again, I think this kind of goes along with regulatory and compliance, Europe in particular is probably a little bit further than we are here in the US in a lot of ways, but let’s talk about ESG initiatives you know, data center footprint or cloud footprint, right? And ESG initiatives are often now getting mentioned in the same breath. Why is that? Tell me a little bit more about that and how we help.
I think that some of the standard bodies out there are recognizing that the demand for data center space is going through the roof, and is gonna continue to operate. And they want those data centers to operate with clean energy. So there’s climate change, there’s all kinds of things going on that’s impacted by all the carbon and stuff that we burn that’s generated by the data centers, which are the biggest culprits. So there is a huge drive for environmental social governance to where they want a hundred percent renewable energy powering these data centers. And all the data centers are moving in that direction. Some are ahead of others obviously. I know the ones that have 100% renewable energy either powered by hydro, solar, wind or even nuclear in some cases which is considered clean energy as well. So it really depends. I think most across the US are probably operating at about 40% clean energy, the rest fossil fuels, carbons, what have you. But I think they’re going in that direction. And if a client is very important to them to get a hundred percent renewable energy, we can help them find the right data centers to fit that need.
Certainly. This has come up a little bit. People in the US see your point, about 40% let’s say, of data centers or users, let’s say in the US are trying to run off a clean, probably a little bit different, higher percentage in Europe and some other parts of the world. Can you share the concept of carbon credits with me? We’ve seen that pop up a lot, whether that be sort of trading credits or gaining credits for going into the right data. For those who aren’t aware, can you just give a quick definition on carbon credits and just sort of that whole sub-market and how that’s affected in and around data center selection?
Yeah, some companies, because they’re publicly traded, because they’re reporting to their shareholders, they wanna be a hundred percent renewable energy, and are only buying data centers that are a hundred percent renewable energy. And what they find when they go to the marketplace is that not everyone has it. And so they might go to, let’s say a data center within the Silicon Valley, and it’s being powered by Silicon Valley Power. And Silicon Valley Power runs about 38 to 40%. It’s published as clean energy. The rest is fossil fuels. And so what you can do is, the data center operator that’s using Silicon Valley Power can get certificates from Silicon Valley Power. Because Silicon Valley Power or actually that data center can go and buy you know, a hundred percent renewable energy credits in another market. Let’s say they happen to be in a market in Las Vegas, and there’s a hundred percent renewable energy there. So they are buying a hundred percent renewable energy, and then they’re taking those credits and they’re issuing it to that customer. So if a customer’s got, call it a megawatt of usage, one megawatt, 40% of that’s clean. So they’re good on 40% of it, they’ll issue them credits for the other 60%. So that client is whole. He goes back to his shareholders, he says, ‘Hey, we’re a hundred percent renewable energy, we are purchasing a hundred percent renewable energy,’ so they’re getting credit for that purchase in Vegas. So it’s kind of being moved around, if you will, in order to satisfy that requirement.
Got it. That’s an interesting concept, right? I mean, it’s almost like it’s gonna become its own currency in a lot of ways, in terms of how we think about that. You know, so hybrid, right? I mean, gosh, word of the day. Everything was cloud before that, everything was as a service. I feel like hybrid’s the new word that we stick in front of everything. Certainly we talk about it in terms of work, but you know, in terms of workloads, it’s a real thing. So talk a little bit about that. You know, I’ll give an anecdotal story and then I’ll sort of ask you to respond to it and sort of give a little bit of background. I was with a customer last week. They buy a lot of other companies, so I’ll just leave it at that. And as they’re doing that, they’re sort of selling off depreciated assets. They’re trying to normalize what the IT infrastructure is, and there’s a lot of old apps sitting out there in data centers on old metal and they know that, despite what they wanna do at the corporate level, there’s no direct path to cloud for them, right? The applications need to be modernized to even be able to go there. But they know that effort’s long and they have to shut down data centers, right? So they’re like, the gear it’s on is fragile, the gear it’s on is old, the gear it’s on is depreciated. And the conversation we had is, what’s the intermediary step? And we started having a conversation about bare metal, obviously. You know, bare metal almost seems like a returning concept from early hosting in a lot of ways, right? But tell us a little bit about bare metal as it is part of the data center story and part of the hybrid story. Where it fits, why people are asking for it, why it’s such a hot topic for us of late.
Yeah. So bare metal is infrastructure as a service essentially without the operating layer. So, a client can get a rack with servers, storage, networking gear, load balancers, whatever they might need, and then just buy it as an opex. And then what these operators have done is gone out and built a nice ecosystem of that. For example, Equinix has 30+ pops of bare metal. So a client can just go on to a website and basically provision over the internet, their own servers, they can even pick the firewalls. They can pick Palo Alto, they can pick whoever they want, and then they build the environment and then they just spin it up and they’re able to start sending data and testing it and using it. So it’s a very easy and nice way for a client to really pull data out of a data center and drop it into an infrastructure that’s being managed for them, and then they can put the operating system on it and then be off and running. So, it’s definitely a nice place. We also talked about compliance. It’s also good for companies or industries that are highly compliant like banking and like healthcare, et cetera. Through those environments they may need to be very careful before they put things into a cloud, and so therefore they might wanna put it into either a private cloud or a bare metal environment.
Decoding Data Center Trends
Got it. So in a lot of ways, think of it like a cloud without the elasticity or ready built private cloud sort of thing. And that’s what made the most sense for this customer we were talking to, because they’re like, these apps need a lot of work before true cloud-based elastic IAS makes sense for them, but they’re like, if we could replicate with modern gear, a similar hardware environment and just stage the boxes as they are to stuff that’s modern, stuff that’s being taken care of by somebody else, and stuff that’s in a data center, they’re not having to retrofit the applications and completely change the way that they’re managing it in the short term while they then take the appropriate time to modernize the applications, obviously. Mel, just one last topic I want to hit, and then I wanna talk a little bit about your framework for a minute. But let me just hit the last topic which is really about the networking attached to this. You know, it used to be that when you were in colo it was like how many racks, how much bandwidth to the outside world, and then how much power. And power’s become, now we talk about colo in terms of megawatts as opposed to in terms of number of cabinets. But I think we also think about the network, the networking differently because it’s not just, ‘Hey, how big is your internet pipe into or colo?’ It’s about the backend fabric of the data center, their ability to get to and interconnect with public clouds logically as opposed to physically. Talk a little bit about how important the kind of the logical networking infrastructure in a colo has moved up the stack in terms of importance as opposed to just the days of wire in the big ole internet pipe into the rack.
Yeah. And again, I think the data center is the central part, right? Whether it’s a core or an edge data center, it needs to be interconnected. Every data center infrastructure decision is coupled with a network infrastructure decision. But to your point Scott, a lot of the larger operators are interconnecting their data centers and they’re coming out with fabrics. These are generally layer two ethernet fabrics that interconnect all of those data centers together. And then they give you an SSDN like portal. So you can go on that SSDN portal and you can just self-provision to anybody they’re connected to. So if someone’s got an ecosystem of hyperscalers, of SaaS providers, of other customers, and you wanna connect to ’em, you just go under their portal and you’re able to provision that on your own. You know, beyond that if a customer needs interconnect data centers, you can always go with a carrier that’s in both of those data centers, because they’re on the net. So just put in a wave, put in a 10 gig, private line, whatever it may be in order to replicate between those two environments. So that is also always a big part of what clients wanna do. But, I think it’s good for all of these data centers to be interconnected because generally they have some data they can share, maybe one backs up the other, depending on the strategy behind what you’re trying to accomplish. But it used to be that everything moved to core data centers and then regional data centers. Now things are more looking to, ‘hey, let’s move some stuff out to the edge to get workloads closer to customers so the customers can have lower latency and a higher quality of experience with that application.’ So, the network is a big part of it. So if you’re with a company that can give you that type of fabric and it’s built in, it’s easy, right? So you also have other companies like Megaport and Packet Fabric trying to accomplish the same thing with an ethernet fabric.
Makes sense. And those companies are in the ecosystems of the data center providers as well, so they can make that happen. Fair to say that that edge case is not just based on low latency on the application side, but also to support emerging IOT applications as well. That’s why some of that logic is getting pushed out to the edge.
That’s true. I agree with that. I haven’t seen the IOT take off like I thought it would. So I think it’s in its early stages, but yes, that’s definitely part of that model.
Fantastic. All right. Well then we’re gonna just wrap up with one thing. Tell us a little bit about how you evaluate an opportunity, without spilling all the popcorn, because we’re gonna have a download at the end of all of this, sort of our framework, but you’ve got sort of 10 considerations that you run through. We got through some of them already, but just without giving all 10, just tell us a little bit about methodically, how we look at these opportunities.
Yeah, so we have the client fill out an Excel spreadsheet with all these questions and understand their current environment, especially if it’s a data center move. Let’s say they’re moving from an on-prem to a co-location facility. We wanna understand the environment, the present state environment. So we have a picture of that with this questionnaire that we go through with them. After that we discuss what their desired future state looks like in terms of hyperscalers, hybrid cloud, who do they need to connect to, where’s their storage, where are their workloads? Understanding that and making sure we consider the 10 strategic considerations that the practice is built on. One of them is low latency to critical workloads. So we wanna make sure we build that in there. Are they deploying a lot of this stuff to SaaS providers? If so, let’s find a low latency to that SaaS provider on one of these networks that’s interconnected with them already. Tax rebates and advantages around renaissance zones. For example, a lot of these people are buying their own gear, putting ’em into a data center. Sometimes it’s tens of millions, sometimes it’s hundreds of millions. In the AI world, it’s billions of dollars people are spending to buy gear. So when you buy gear, if you’re in a data center in California, when you ship that gear to that California data center, you’re gonna pay close to 10% tax on that gear. So 10 million bucks is a million dollars of taxes. That’s material money, so if you pivot, and if you were to move that to Hillsborough, which has a renaissance zone or Arizona, you pay zero tax, so you just saved a million dollars that needs to go into the ROI. So we basically help them understand those things. It might be important to them, it might not, but we certainly can help ’em with that. We talked about green energy, ESG initiatives, PUE is something we talk about. Power utilization effectiveness. How efficient is that data center, right? A low PUE, like 1.3 for example, means that for every kilowatt that a data center operator generates, it takes them a third or three tenths of a kilowatt 1.3 to cool it. That’s the overall cost. So that also means it’s an efficient facility. If they’re in a 1.6 or a higher PUE facility, their fans are gonna kick on all the time. The gear is going to be taxed and their refresh might not come in five years, it will come in three because the gear has been taxed, because they’re in an inefficient environment. So those things are things we talk about to make sure they understand. Care availability, peering exchanges, those things are also important. Hyperscalers they wanna connect to, how do they connect to hyperscalers, the network piece you talked about. So those are just a few of the things we talked about with the client and some of ’em are important to ’em and some of them are more important to them.
Expert Strategies and Insights for Future Moves
So Mel, let’s wrap it up with success. I mean, and I’m gonna put you on the spot on a number. I think the last time we were able to calculate this, where we had numbers to compare it to, it might’ve been 21 or 22, I’m not sure. But you know, I’ve heard you share a stat about the amount of kilowatts of colo that were sold in a particular year in the United States, and the amount of that that was consulted on, sold through, if you will, your data center practice, and obviously the entire broader bridgepoint team. Can you share that?
Sure, sure. I’m only sharing the one that I’m definitive of. And it was actually four years ago when we launched the practice in 2020. That year, Jll, which is a real estate broker, came out with a report that said 680 megawatts was procured in North America of data centers. Capacity our practice had been launched that year. We did over 70 megawatts of non-hyper scale, which represents around 14,000 cabinets that came through our practice. This is pre-COVID or just right around COVID. So that represented about 11% of the market that we were a part of. I can tell you since then, it’s gone north and I can tell you that capacities that people are buying today are, you know, not 680 megawatts. It’s over two gigawatts. So it’s gone up quite a bit.
That’s amazing. That’s amazing. And evidence as to why, and as is this whole conversation, frankly, evidence as to why customers out there in the field should be looking to Bridgepointe for help basically in concierge, there I don’t know if that’s a word or not, but I just made it up. Their data center decisions you know, more knowledge, more vendors you know, and we’ve just done this 14,000 cabinets strong, I think was what you said four years ago, and the numbers have only gone north. This is what we do. Head of the practice sitting in front of you. Mel, we’re gonna do this again. I gotta get at that big brain a few more times, but I was super happy to have you on you know, this quick hit episode of The Bridge and thrilled that you were able to be here with us today.
Thank you, Scott. I appreciate being here. Thank you so much.
Absolutely. If you wanna learn more about Bridgepointe’s data center practice, we’ll include in the show notes a link to the data center practice on our website. That’s also where you’ll be able to get access to some of the content that we’re putting together with Mel around the 10 important considerations that you should be thinking about when you’re planning your next data center move. And we’ll also include in the show notes a link to Mel’s personal page where we’ve got other blog articles and things that he’s been involved in. So Mel, you’re a star. We were thrilled to have you here. And to all of you out there, we will catch you soon on another episode of The Bridge.