 Live from the JSA Podcast Studio, presenting Data Movers, showcasing the leaders behind the headlines in the telecom and data center infrastructure industry. Hey everybody, welcome to a special edition of our Data Movers podcast, The Greener Data Edition. I'm your host, Jamie Scato-Cutia, CEO and founder of JSA. And with me, of course, my fabulous co-host, top B2B social influencer, Evan Christel. Hey everyone, and welcome to Data Movers, where we sit down with the most influential men and women in today's data center and telco world, supporting the network infrastructure requirements of our modern world. But before we get all techy, Jamie, did you want to see Oscars recently? I did, I caught bits and pieces. I was working on a proposal. I think there was one scene probably you recall, catching involving, you know, some slapping going on. I consider you kind of a Hollywood type. And so very, very tuned into what's going on in Hollywood and with stars and that kind of thing. But you're also an expert in crisis communications, crisis management, you know, PRs falls under PR. What do you think could have been done to handle that whole episode differently on the communication side? Well, obviously, you know, hey, for ABC, it was great. They got the ratings they wanted. But the slap heard round the world is the headline that was interesting. But, you know, from a PR perspective, ideally, you know, you don't do that. Like, just try to avoid physical altercations whenever possible. And, you know, it's so sad because, you know, really that night when he won Best Actor, like that was his night that he earned for so long with his beautiful career and instead, the attention just got, you know, on that action that I, you know, you can hear in his apology statements thereafter, you know, that he has remorse and, you know, he apologized even during his acceptance speech. When, you know, when we just do things out of gut response and, you know, he's protecting his wife, which is hard, it is endearing, but obviously violence is just never the answer. So from a crisis communications, stop what you're doing. Apologize, apologize from the heart and hope that the music, that the news cycle changes quickly. And that's a good advice, right? Yeah, and I was amazed that no security was there to kind of stop them from coming on stage or even to step up afterwards. It was all very strange, bizarre. I think anyone else, but it was Will Smith coming on to stage and everyone felt, you know, and it got a lot of folks in their interviews, post-event saying, we thought it was scripted. We thought this was like, you know, part of the stick. So I think people just were all sort of caught off guard, you know, and I know in being in newsrooms and broadcast rooms is usually like a red button that we press when, you know, to stop the broadcast, it's live, 20 second delay, and I'm sure ABC was like, ah, stop, stop, stop. No red button. Well, we don't have a red button either, but we're much kinder and gentler here on Data Movers. So we don't subscribe to any of this, but yeah, let's get on with our guest. Yes, oh my goodness. And one exciting guest we have, as always, you know, this is the podcast where we really sit down and we dive into the background stories, career highs and lows, unique perspectives of these industry leaders. And today, no exception. We are so excited to welcome Karim Sheikh. He's the CTO at Virtual Power Systems. He's also a faculty member of the Integrated Innovation Institute at Carnegie Mellon University. Karim is also, and I'm so proud to say, contributing author to our book, Greener Data, making its debut on Amazon on Earth Day, April 22nd, just 24 days away. Welcome, Karim. Oh, thank you. Thank you, Ivan. Thank you, Jimmy, for having me on this Data Movers podcast. I will try to not create any Oscars moment here. Okay, well, I think we won't go there. But before we jump into an introduction to yourself and Virtual Power Systems, BPS, I was intrigued. Not only did you graduate from an amazing institution of PhD at Carnegie Mellon, of course, IIT back in India, but you also teach there today. So you liked it so much, you just had to stick around and teach virtually. Yeah, yeah, sometimes, you know, you just can't get away from your roots that far, right? So I actually, there's a campus in Silicon Valley that I teach physically yet. So it's actually not virtual. So I actually get to engage with the campus and the students. So it's a lot of fun. It sounds like it. And trust me, you won't see Jamie and I in any of those classes where it's a little bit above our pay grade. But how did that lead to Virtual Power Systems? Give us some background into your journey and today's role at BPS, which is really revolutionizing the power infrastructure management side of the data center, which is really top of mind to everyone in our industry. Yeah, actually, so it goes back to when I was in India like almost 40 years ago, that kind of puts a date on my me, but you know, I think it was great. It was really great that I was fortunate and grateful to, at 40 years ago, to be able to use technology and computers in India. It was just coming in, but I just happened to be at IIT where I had access to it. So I developed a passion for both computers and software, and it continued on in my career. And eventually I ended up at BPS where, BPS really started off with the idea of using batteries in racks to help balance power consumption and kind of increase the just utilization within the racks. That was like eight years ago. Over the years, we really learned a lot more about the industry. And so we ended up focusing more on software and less on hardware and really looking at how software can dramatically change the way power is utilized in the data center and increase the power utilization across the entire infrastructure rather than focusing on a rack. So I'm really looking forward to sharing that idea and that approach in our podcast today as well as in the chapter that I'm writing for the book on greener data. Yeah, and that's actually a great transition because as you stated, you've been immersed in technology from an early age. Can you tell us what fascinated you about the industry? What led you to change your way of thinking about power infrastructure, especially when you met Shanke Ramu Muti, the founder of BPS in 2014? Sure, Jamie. It's a great question because it almost happened by chance, really. So I've been building software applications all my career, deploying them in data centers. Never really thought twice about the facility side of the data center. For me, my focus really stopped with the IT servers. So for me, the rest of it was not something that I really paid a lot of attention to. So a mutual friend of mine introduced me to Shankar and I really wasn't sure what he was trying to accomplish because to me it sounded like a boring infrastructure problem. But as he really started out articulating to me and telling me about how data centers are very inefficient, and it also translated to this kind of impact it has on the environment, it seemed like the problem is much bigger and there were aspects of it that the industry was really not addressing. So BPS really had a great opportunity and a good position to look at sustainability from these angles that the industry was not looking at. And that's that way. I also realized in the process that how big of a problem it was and it's kind of hard to ignore once you realize it. And if you think about it, infrastructure measurements estimates right now that up to 37 gigawatts or maybe even up to 50 gigawatts of power infrastructure is actually built but not utilized. So that's a powerful number. So it was a very compelling opportunity for me to contribute and make a difference in the space. And so I decided I'm gonna jump on to what BPS is doing and move that product forward. Yeah, the push towards a sustainable data center has been in action over the past couple of decades but really seems to be intensifying now with all the green initiatives under foot. So I'll just understand why power infrastructure is underutilized that he described. And yeah, my electrical engineering mind, I'm trying to reconcile two different problems, software and hardware. And how do you bring intelligence and resilience into the power world? Great question. Yeah, that's a really good question. For this, I'm gonna go back to about 2006 when the data center industry adopted this power usage effectiveness PUE as a very important measure of how much of the power in the data center actually goes to IT loads. And that was a good way for them to start improving on there was something you could measure and do something with, right? And the value went, I mean, it started off with around 2.0 or greater than 2.0 and over the years, it has actually come down to in many data centers to less than 1.1. That's a tremendous improvement, right? But it's also interesting to note that most of this was achieved through improvements in either the mechanical systems or the electrical systems. And so there was a, for example, putting in more efficient cooling technology or improving the space design within the data center to get better flow of air and to get the better thermal flow in it. So things like that. The one thing that the PUE does not really address is what we call a stranded power capacity in the data center. So that's really what we focused on referring to the question we're asking. So we define any stranded capacity as underutilization of a power infrastructure that you've already built, right? So there are three reasons, particularly that it happens and that we are focused on. One is that data centers typically have redundant infrastructure. Then they will put additional hardware, extra hardware, mainly to account for potential failures in some components of the system. So these components, however, if you don't have a failure can go the entire year completely unutilized, but there's just sitting there. So, but it's important because they help data center be reliable. So you can't really say they're not serving your purpose, but they're not used. Second, the data center operators also allocate less capacity than their infrastructure to tenants. And that's because they want to be safe and they want to protect themselves from unexpected spikes and peaks that some tenants may bring, especially something that may trip their breakers. And you don't want something one tenant does to impact another tenant. So you just give yourself some wiggle room so to speak, we call them the buffers. So this is another reason why you will end up actually using less of what the infrastructure that you've built. This allocated capacity is the actual capacity they sell to, they contract with tenants. The tenants, when they contract with the data center, they don't really, they try to match their contract to the worst case utilization that they expect to have, which means majority of the year, probably 95, 99% of the time, they are significantly under that number. So there's another place where you have underutilization, but they can't afford to not contract for the full amount because they will need it on that Black Friday situation. So that's what happens then is you now have underutilization within tenants as well. So when you combine these three, so the redundancy, the buffers and the underutilization by the tenant together, we estimate that it is about 60% of the infrastructure that's actually in the data center, that's not being used. So if you go back to PUE that we were talking about earlier, if we address this part of the problem, I believe that we're gonna pick up where PUE is reaching its limitation on what it can do with the mechanical and electrical components and gives us a new area where we need to develop innovations on how we can better the infrastructure. And I think that will be an additional contribution to the sustainability efforts that we need to take in the industry. Yeah, and I love this idea too. I mean, that's 60% of stranded capacity that we're not utilizing, that can mean a meaningful difference. And I know VPS has been working to really bring power virtualization to the mainstream for over eight years. And this idea that you touch upon in your chapter and greener data, but the idea of autonomous digital infrastructure and software defined power now being really a key differentiator to data center operators, can you give us a little bit more? Oh, sure, yeah. I mean, that's really the focus of, the message we want to communicate is this idea of how we can use software. So I want to talk definitely more about it. So I want to just relate it back to again, the idea of the standard capacity we just talked about. So the point of what we are trying to do is that we believe that the standard capacity can actually be unlocked as new capacity in the data center. Just to think about it like, we do want to think a little bit about why it exists in the first place before we jump into how we're going to use it. It sits there because it helps the data centers be more reliable, it reduces the risk. And most importantly, it gives peace of mind to the data center operators that there's less likely for something to go wrong. Since a lot of response to any events in the data center is still very manual in the data center today, it makes sense that we have so much buffer because the human reaction time is low. So you can't react immediately to when something happens. But if you look at other industries, software plays a key roles in many mission critical systems, right? We use it for launching astronauts into space, for managing air traffic, and for everyone knows autonomous vehicles. So is there an opportunity for us to leverage software in helping us out here in achieving this unlocking the standard capacity? So I'm gonna say no, we haven't really done enough in terms of using software for all that it can do in the data center. So what we, our, I believe is that with software and its ability to monitor and respond to situations in real time, we can actually unlock quite a big chunk of the standard capacity because now we are matching our ability to react, the speed at which we can react with how much we are able to leverage the power infrastructure and reduce this buffers and redundancy requirements. So I think that's really the approach we took when we started working on our platform is how do we, the more automation we do, the more quickly we can respond, we believe we can unlock more and more of this capacity that's there. And 60%, I mean, that's meaningful. So instead of us feeling like they have to build new, new, new all the time, we can actually utilize more of what we have. Oh, absolutely, yes. And the number is also important because the first 10, 20, 30% of that 60% can probably be unlocked with very little concern. It'll adjust initially. And then the more sophisticated the software gets and its ability to orchestrate the data center infrastructure, you can unlock more and more of that 60%, right? So we just wanna start with the first, whatever the first 10, 20% what we are able to actually unlock is that's a good start. That's already making a difference. So what we, what I'm trying to do in this chapter is really draw a comparison between the IT industry from 20 years ago where they resisted virtualization as a risk, the software defined compute what VMware was doing. And now it's widely used and it's like, it's an automatic decision. Everybody uses virtual machines when they're building out software now, which was unthinkable 20 years ago. That's called software defined compute. And what we're doing here is the power equivalent software defined power and how it enables sort of an autonomous digital infrastructure where software doesn't just make decisions but it also takes actions based on those decisions by interacting with the hardware. So even you talked earlier about the software and hardware and how it like bringing that idea of them working together that's what we're trying to do here is work with hardware vendors in the industry to say, hey, there is something you can do with your hardware that makes it do actions. And we are able to control them and have them take actions at certain times to manage the risk versus the, how much of the capacity you're able to sell and generate revenue and make your data center more utilized. So for me, the top, really the top takeaway from the chapter would be for the industry to see the potential of automation in the data center through software and how it can change both the economics and the sustainability aspect of their operations. So for some, I think it is definitely a pattern shift worth exploring, I think. Absolutely. Amazing. We've heard the old adage software eating the world. Now we're seeing software eat the data center and the power that powers the data center. So it's really intriguing. Now, what would you recommend any practical advice to data center designers or operators to help them rethink how they build design, future proof their data center and their contracts from a sustainability standpoint? Yeah, that's a great question, Devan, because I think what I've said so far is easier said than done in an industry that mostly has been able to survive without any fundamental changes to its business model. Because something there is gonna have to change to really have this be effective. So I think getting the most out of software, different power requires the stakeholders in the industry to make some adjustments to their business models about how they do contracts. Mainly they need to be able to support new product offerings that are more consistent with tenants individual needs rather than saying that this is the industry standard product and that's what everybody is going to get. So we're gonna be asking that, you know, rethink and get more alignment with your tenants. So in this chapter, I actually draw a parallel in with how cloud service providers like AWS have evolved their business model from simple offering a very simple virtualization to not a choice of products that are tailored to the market needs and each customer is able to log in and choose exactly what they need. Do I want reserved instances? Do I want spot instances? What is my needs in terms of capacity? So same way that they did in IT, we can do that in power. So in software, different power, the data center operators can offer different options with different SLAs and really giving them what we call as the right sized contracts that adapts to their needs. And at the same time, some contracts with multiple tenants with a mixture of SLAs and get creative on it just like AWS does for their cloud services to really utilize the full capacity of the infrastructure in some cases multiple times over. And so I think that's the thing. The software is aware of the SLAs and ensure that we can actually put constraints in place to make sure it doesn't let you sign contracts or a combination of SLAs that are actually not easy to accomplish so they can help you with that. And once you sign the contracts and the data center, the contracts are in operation, it'll ensure that when events occur, we orchestrate the workloads to comply with the SLAs that you have with each of them. And I think talking about data center designers, there's another thing that I think an interesting opportunity for them while we can certainly go back and have the existing data centers leverage software defined power, I think the really interesting approach is what does it mean for new data centers that haven't been built yet, that are being designed now? So they can take this into account and start designing with a full advantage of a highly flexible data center that can adapt to a number of different SLA requirements easily and dynamically. So I think that's a great opportunity for them. So I think my hope is that eventually all data centers are built with sort of this flexibility baked into their core design and you're offering a variety of SLA products right from the beginning, your customers, right size to their needs as opposed to what we think is the industry standard or what we believe everyone should have. Brilliant, brilliant. Of course, Karima, of course you are an industry thought leader and so I know this next question is super hard but I gotta ask, where do you see our industry in 10 years from now? Yeah, I think, you know I wish I could predict everything but I would say, you know one thing I really like is innovations are continuing to happen in the spaces that have happened like the mechanical and electrical space I think that trend will continue. I do expect that there will be some really new ways that we take towards the greener data. And I, you know, I think one thing is taking software as an example I think though we are doing something interesting in the video, I think we are just addressing a piece of it. I think there's a lot more that can be done. The management of the infrastructure can go very granular, can be more creative. So I think this, and we can actually eventually get to maybe a 100% utilization. And I think the bigger innovation is going to come at what I believe is the intersection of software and hardware. Right now we are pushing a little bit in the industry but at some point the hardware vendors are going to realize the value of it and they're going to make their devices and more intelligent and in a way really create a chemistry between the software and hardware that really gets us to 100% unlocking of the infrastructure, I think. What I really like that's happening is the fact that we are starting to really think about because you can only make improvements if you can measure it. So what I really like is the, what's going on with the infrastructure member. So let's do this one. With the infrastructure members where they're looking at this way of carbon sticker that goes with every data center. And if you're able to identify the cost of every network packet that's going through a data center and then you connect it across all the data so that it went through, you know the cost of the data that got to you in terms of its greenness. So I think that's a, I really like the fact that we are focusing on measurement of something concrete because we can now try to figure out how to improve on it. So I think there's a lot, I'm really excited. I don't know where this is going to go but I really am excited about how we're going to get to this idea of a really sustainable data center as we go forward. You know, that's amazing to hear but I was really hoping that in 10 years you were going to talk about data centers in space and on Mars and on the moon. So I guess I'll just have to wait another 10 years to hear that prediction. Probably, but I wouldn't be surprised if it happens. Yeah, yeah. Okay, well, let's wait for that. In the meantime, we're going to do some fun rapid fire questions to try to embarrass you. I mean, try to get some insight into you as a person. So let's start off with your motto. Do you have a motto? Can you think of a motto for everyone to follow personally, professionally? What would it be and why? Think of one, that's always hard, right? But I'd say if I had to think of one, it would be to stop and look behind you because that's your legacy. And I say, big or small, make sure it's positive, right? And I think because I think we're all so focused on our goals and the momentum forward all the time, right? That we don't really pause and evaluate and say, okay, what impact have we had on our family, on our friends and the world, from what we have done in the past? I think there's a lot to learn from that too and say, okay, how do I make adjustments to really make sure it's something positive? Because we all don't need to leave a big legacy, but I think we need to leave a positive legacy. Wow, that's deep. Thank you. Yeah, I love that. And really the genesis for greener data. So really well in beam, my friend. And of course, I'll know you as an inspirational educator. What do you enjoy educating others most about? Yeah, first I wanna talk about what I learn from teaching because that's always interesting, right? So, and I think one of the things I really am enjoying is I get to learn about this new generation of people that's coming into the world before they actually come into the industry. So it's kind of interesting to watch them say, okay, these are the people that are gonna come in and do what we have all been doing, right? That's incredible to get to understand their mindset. So I think that's been really great. But for me, what I try to do with teaching with the courses that I teach is I try to open the minds of young students, especially because I teach students who are in the technology field. I try to expand their mind outside of technology and into awareness of social impact of their decisions. So one of my favorite topics is we teach our futures thinking where students actually imagine a future that you cannot logically get to from the current state, right? And so this makes them think of ideas that wouldn't be possible if they were only interested on incremental improvements. So it's actually really exciting to watch them just being that space, which is actually uncomfortable, but they really enjoyed and the creative ideas that come from it. So I think that kind of is really my favorite topic and these are students actually going to product management in the industry. So I'm hoping that they carry some of that type of thinking into the industry and actually use them in their product strategy and ideas. Love it. Thinking about sustainability, what's one thing you're trying at home to lead a more sustainable lifestyle? Side note, I tried composting once, it's a complete disaster, mess, smelly, it was horrible, but it didn't stop me. So still trying, but what about you? What are you trying to do at home conservation wise? So my wife and I do gardening. She is an avid gardener. So I can say I'm an assistant gardener, but I think to my wife, she'll say, I'm a gardener trainee, more than that. But that's a lot of fun. But I think the easiest thing that I've done for my part and I think many people can do is like just having solar at home and I have installed a battery. And I think what I really like about that is it gives my family peace of mind for me a cool toy to play with. And less energy burden on the world. That's amazing. Did you go with Elon Musk's solar scheme or did you do it yourself? I must admit, I went with Elon Musk's scheme. Hi, Dean. Oh, good for you. That's an endorsement. That's great. All right, so what is your favorite outdoor activity? So actually we live in the foothills of the Santa Cruz Mountains here. So we go hiking. Okay, rub it into less people in snow here, but that's okay. We'll live vicariously through you. Yeah, so this is California. We get few days of bad weather. And I have to admit though, I'm not very proactive about the hiking, but I absolutely enjoy the outdoor feelings because that goes back to where I grew up. And it's definitely a distressing effect on me. That seems to be pretty much needed at times. Absolutely, especially these days. Well, thank you so much for joining us, Karim. It was amazing to hear how software defined power infrastructure is even a thing. I had no idea. And we will be watching. Yeah, it was great. Thank you so much for having me here. I enjoyed the conversation and the opportunity for me to share what I'm doing. And so I really want to thank you for taking the time and including me in this. We are so honored and blessed to have you here on Data Movers and in greener data. Again, do out Earth Day appropriately, April 22nd. And if you enjoyed guys listening to this podcast, I hope you did, please be sure to check out jsa.net slash podcast for upcoming Data Movers episodes releasing every other Wednesday morning. And Evan. Until then, follow us on social at Evan Kerstel and JS Goddow, and we will tweet back. Yeah, and guys, stay safe out there. Think green and as always, happy networking.