Conversation with Antonio Pellegrino
Good day everyone, and welcome to the latest Momenta Podcast. My name is Leif Eriksen, Insights Partner here at Momenta, and our guest today is Antonio Pellegrino, Founder and CEO of Mutable. Pele as he’s known has been involved in several ventures since co-founding his first computer services company at the of sixteen. Recognizing the central role that Edge computing would play in 5G, he launched Mutable in 2017.
Hey, how’s it going?
Let’s start today talking a little bit about how you got into computing in general, and cloud computing more specifically. Tell us a little bit about your background and how you got to where you are today.
I started out working in coding quite early, so in that regard I actually did everything in High School, freshman, everything from robotics to gaming, and building a streaming site for gaming. Creating EUF which was a broadcasting company for eSports during 2007 all the way up to 2011. So, it was really an immense effort with 30 people throughout the world, and in that regard essentially working with servers and everything back then, I kept on feeling the need myself of basically creating tools to help build all the ideas that were in my head. In that regard I ended up shifting from streaming video, over to developer tools, making tools to help me build my own things that I’d been building, and then all these companies around me going through NYU, the engineering School, they all need help building what they were doing as well, and create a company focused on more or less the cloud.
In that process we realized that one of the customers was really geared towards wireless charging for cars, as crazy as that sounds, and in order to pull that off they need to compute everywhere, and we wanted to make sure we were able to do that. So that’s when we created Mutable in order to focus on trying to find compute everywhere, so you can run applications everywhere, but not have to actually physically have them everywhere yourself. So, create like an Airbnb for servers, making a Public Edge Cloud.
Interesting stuff, and quite a journey. So how does your experience today inform your views on cloud computing and where it’s going?
Well by having these differences of backgrounds from robotics to streaming video, and IoT, it really made a lot of sense of what I was doing, to figure out how do I take the knowledge of pain-points that I had during those times, and apply it to all these companies that are sprouting up to solve these problems like AR/VR, gaming, and all this stuff. I realized that in order to pull that off, what needed to happen was compute more locally, and that’s where we really made that shift to go in that direction.
And you refer to this as Public Edge Cloud, so maybe talk a little bit about the difference between say Public Edge Cloud, and Public Cloud in general, which everybody knows about, but what is Public Edge Cloud, and how is it different?
We had our own identifier for Edge, and the reason for it is, in the public cloud it is the idea of using other people’s servers, and for the most part that is mainly the three hyper-scalers, its Amazon, Google, Microsoft, and they’re centralized in basically different parts of the world, but we’re talking usually about usually hundreds, if not thousands of miles away from wherever the people are, and devices that are using it, versus, when we talk about the edge, we’re talk about the network edge, we’re talking about cable companies, Telco companies, we’re talking about 25 miles away from your home. Think of it like you’re doing Amazon and you’re buying something, and you’re getting it from a warehouse, versus going to your corner store at a CBS and being able to pick something up right there when you need it. That’s really what we’re trying to push forward.
And so, who needs this? Lots of organizations, businesses, individuals are accessing data from these cloud services, from AWS, Google, and Microsoft. Why does anybody need this edge cloud service, and what kinds of organizations need it?
What we realized is, this is really about three pillars, things that really focus on latency, trying to bring things snappier and faster, so for example, gaming, anything to do with IoT where you’re basically collecting massive amounts of data, and having to have something sift through the data, get a response right away and have a control loop; so think of a factory, think of drones going around, all these types of things, you have this constant feedback loop.
Bandwidth, so think of recording video everywhere, let’s say you’re in a factory and you have a whole bunch of cameras looking around, in order to judge what’s going on, or outside at an oil pipeline or whatever it may be. Basically, taking that video feed, uploading it, and doing imagery recognition right away, all these types of things require a massive amount of bandwidth. Transporting that all the way to the cloud, that costs a lot of money, it also takes too long to be actionable.
And then the last being security. When we talk about security, we’re talking about trying to do things before it ever reaching the internet. So, what that means is, if you have cameras and devices, and everything out in the field, or inside of buildings, and then you have transporting it all the way to the cloud, usually that means you’re going over the general internet. The internet can be a scary thing, its not a straight line going from you to a massive datacenter, but it is a straight line typically to go from you to your internet provider. Being able to run this style compute inside of that facility allows you so that it doesn’t reach the internet, its inside the safe firewall that more or less you control, and that’s really a lot of what we tend to work with on the security side.
Interesting. You mentioned the term ‘latency’, where do we talk in terms of differences, I guess in response time or whatever, between public edge cloud and public cloud, what kind of numbers are we talking about?
Well the numbers we’re talking about is the public cloud, what you typically have these big hyperscale datacenters, is usually around 50 to 100 milliseconds. Whilst when we’re talking about the public edge cloud, we can go down to 5, 10, even sometimes 1 millisecond, and having compute literally right there. So, it’s those concepts of data gravity, ‘Don’t move the data, how about you move the compute towards the data?’ It’s a simple pattern of that.
And it’s already there right? You’re talking about leveraging existing compute resources that are there and available at certain times, is that correct?
Yes, that’s the whole concept of Airbnb for servers. You have to think that all these companies have compute out there for a particular reason in all these regions, purposely for themselves. They have it for cable and wireless companies, they have it for video streaming which is mainly from 8pm to 10pm. So, they have all these servers just like Amazon did, for three days out of the whole year, and then they created AWS, Amazon Web Services out of that because they’re like, ‘What are we going to do with all these computes sitting here most of the day?’ So, we’re doing that across the board, across all these companies that are out there, that have servers miles away from where you are, and where the data is, and by collecting it all together and creating one cloud out of it, that’s where we’re able to make the best of both worlds. Not having to physically have CAPEX everywhere, but allowing people who already have that CAPEX deployed, will allow them to make money in addition to create a secure and more potent environment for them to run applications in.
That makes perfect sense. But let me take it in a slightly different direction, in the sense that it’s one thing to provide that low latency for folks that are doing gaming, which of course they need that as well but, businesses who might want to take advantage of that low latency, and even industrial organizations, governments etc. they might want to make sure that it computes, and the link to that compute is very secure, and might shy away from the term public edge cloud. Is there a private version, or is this just a terminology issue? For organizations that would want to say, ‘Well, I need to have all this stuff walled-off for my purposes, and never to be touched or penetrable by anybody else’s’. How does that fit into this equation?
Well, actually what has been happening in the virtualization space of servers has gone an incredibly long way. So, what the AWS guys have created, and Google, for security in isolation, is quite great and we’re bringing that over to you, instead of you bring everything to them. So, essentially what we’re doing is, we’re providing the same environment that AWS would create, where everything is walled-off on the virtualization layer, as well as on the network side, we’re doing hard IP prefix firewall on everything else. So, everything is quite secure on this compute, and that’s really what our huge focus is on, because for us being able to take over this compute and run multiple things at the same time, we have no choice but to make sure that everything is as secure as possible.
So that’s something that we focus on, and we really make sure that is isolated as much as possible, so that you are provided the best experience. And it’s not just North and South being firewall from data coming in on the end-point side, and then on the internet side, we actually make sure that each individual application running side-by-side on the same box is completely isolated, and that’s really important.
Great clarification. And so, is there anything else with respect to cyber-security that separates or distinguishes public edge cloud, versus just public cloud?
The other big thing is, if we’re doing compute inside of your environment, especially if an internet provider is providing your SD-WAN, which is your software defined networking and has the global firewall rules for the whole company, and we’re inside of it, we’re able to make sure that everything is completely ruled off based off of your standards, and it’s essentially on-prem without having to de-unprem, so you get all the benefits of the security of being on-prem, with having the compute that’s right nearby. So, by working with the cable and telecom companies, and the private networking companies, there’s a bunch of those spinning up where you can create your own private LTE, and 5G and all that kind of stuff, we’re definitely able to provide that kind of thing on top, in order to [lost audio 14:03].
That’s interesting, and so there’s obviously a whole variety of different applications that could use this service, or this approach if you will, to cloud computing.
What would you say Pele is the role of the communication service providers in all of this, the telco’s and the cable companies etc. in making this happen and growing this concept of public edge cloud?
What’s great about the cable and telco companies, is that these guys have hard assets that need to be in certain locations, to provide a guaranteed quality of service. So, they’re purposely putting extra redundant compute out there, for their own purposes. So they have to spend all this CAPEX, and if we can change that CAPEX, instead of it just being something that is lying dormant in a cost center, and changing that to revenue generation, that helps everyone, because that means the telco’s and the cable companies can now make money off of that, and not have to pass that cost burden over to their customers. In addition, the customers now have a source of compute that is literally [audio breaks 15:38] can access to provide them a better quality of service. We see the cable and telco companies as the ushering in for the edge compute, by them having this real estate and compute or existing.
So, as you said, Airbnb, Uber, etc. for cloud compute in that sense, very interesting.
I want to come back to the characteristics of an organization that would benefit from public cloud compute, again on the end-user side, because this is something that our listeners would start to be thinking about, ‘Is this something that I can use?’ ‘How can I use it?’ So, from your perspective, if you were to breakdown the characteristics of an organization that could use it, or even needs it, to run their applications, help run their business, what would that be?
Think of things that typically are running on device or on-premise, that are either wickedly expensive, or need to have certain parameters in order to be performing enough, and that’s the reason why you did it in the first place. Things that are running that are secure, that are meant to be internal, all those types of things can now get all the benefits of the cloud, without actually being thousands of miles away. Think of as I was saying before, cameras that you already now have in your businesses, that can be converted into smart cameras, and that mean they can see what parking spots are available, see things that are supposed to be monitored 24/7, and if there’s any change in what’s going on, instantly being able to monitor that change, report it and sometimes have something automatically go there to check it out.
Doors left open, air conditioning, all those types of things, for security purposes make sure that’s happening, just from looking at cameras, tracking of assets and people. You also can imagine fleets, one example as I was saying before is a customer of ours is wireless charging for cars, and when you think of that what that really is, is essentially a charging pad with a car driving on top of it, and what you need the real time or snappiness for is the alignment, you have a car and a person in it trying to react to driving, being exact on top of this pad in real time. Imagine having to do that with a bad connection, just going back and forth, back and forth.
Yes, the latency of that.
Exactly, exactly. That also applies where they have infrared cameras looking at this, so that way if there’s any built-up heat from metal being on top of it, or a human going underneath it, or an animal, they can react to it in real time. So, all these types of things where instead of putting all the smarts inside the device, you make the device as a sensor collector, and then everything else is done in managing the cloud, or the near-by cloud, in this case the edge cloud, and that really reduces the price of that device. So, instead of having a $5000 camera, you can have these $25-$50 cameras and have 100 of them provide the same or better actions with a lot cheaper job cost.
You know what that brings to mind to me is, government and public safety and things like emergency response, dealing with natural disasters, you’ve got to capture a lot of information, and get a lot of information at the point of activity, which is always challenging. In those scenarios you hear about networks going down, communication, not just verbal communication, but that means data communication. Do you think this is something that can be useful if that regard, in those kinds of scenarios?
Yes, one of the actual great use cases of all this is, you have to think a lot of this world is becoming machine-sucking-machines, less of humans telling machines what to do. Right now, we’re at the speed of humans telling machines what to do, and in this case be able to make them just freely talk to each other and provide us connectivity means network monitoring, identifying threats and acting on it right away, and that could be in the network, that could be data threats.
For example, there’s initiative right now of monitoring the electrical grid. If you monitor the electrical grid you will actually see in real time all these natural disasters come through, tornados or whatever else, and see how things go down and come back up, and constantly target where or secure where you predict the next event is going to happen, as storms roll through. So, all these things become very important in planning and reacting in real time, versus you get a recording after the fact and being like, ‘Oh, okay that’s what happened’. It’s like catching someone in the act, versus ‘Oh that guy’s wearing a hoody, I couldn’t figure out who he was, so oops, that recording is not very helpful’.
Exactly. You mentioned the grid, of course, the electrical distribution world is moving more towards a distributed grid, where who as producer of electricity versus who as a consumer is going to blur a little bit. You mentioned again electric vehicles, at night they may be inclined they’re not going to use the car the next day, to put some of that power onto the grid. People with solar panels, I could see this also being useful in that context as well, because there’s a lot of information exchange that has to happen as part of that brokering of where power is produced, where its consumed, and how that whole thing is managed.
What’s the future of this technology Pele, where are we going from here? It’s early days right in terms of the use of it, and the opportunities, where do you see the future of this going? Are there technology changes for example coming down the pipe, that are not here yet but will impact it?
As I was leaning towards earlier, the way that I see this is, we end up in a world where we all carry around as a glass plate, everything is streamed to us, and essentially everything is interchangeable, but personalized; from being in an airport and walking down the hall, everything pops up is related to you, to essentially the environment around you reacting to what’s going on. So, what that means is, just a simple thing like VR, we see VR as this futuristic thing that we’ve been trying to pull off for the last 30 years or more, and it still isn’t there yet, and why? The main reason why is because of, how much will it cost for anyone to get access to it? You have to be tethered around with the computer in order to pull it off, to have a good quality experience.
So, if we can actually run the VR experience off onto these servers nearby, you can actually have this experience where essentially you get maybe a $300 dollar pair of glasses, and you can use them anywhere, and no matter where you are, over wireless, over Wi-Fi. Being able to have that experience that changes the whole paradigm of being more like a Game Boy, where you can use it anywhere you need, versus something that’s like an appliance, it’s kind of fixed in your home.
The reason why I bring up VR is because those are tools that are now being used for training everywhere, from simulations for the military, to retail source, where they’re trying to train their representatives on how to react to customers, and do customer training, or working with industrial equipment before breaking it.
AR is another one of those augmented reality, where you’re wearing either glasses or having it on your phone just looking at the screen, doing image recognition and applying content or things on top, that’s another huge part of what’s happening in the industrial world, to take advantage of what should be taken off the conveyor belt, or what shouldn’t be; what’s the proper way of turning on this machine or not? All these things are now becoming assistance where the individual doesn’t have to have that knowledge, it can be applied on top of the real world.
That’s the stuff of a lot of the ads we see that we know are not reality as far as VR and AR, but yes, it’s a conceivable future. When we talk about machines, there’s not really any privacy concerns per se. You started by talking about people walking down the street, through an airport or whatever, things popping up based on them, of course that raises the thing that everybody talks about this day which is data privacy. What are your views on that, and how it’s going to impact how this all evolves?
Privacy is interesting because we’re starting this world of GDPR and the California regulation which is going to expand throughout the US and Canada, I’m sure. All this is about is data sovereignty, meaning you own this data, and the data is controlled and managed by you, versus a third party accessing it, managing it, keeping that and doing whatever they want with it. So, the way you have the sovereignty is to keep the data close and to keep it managed by you, so that’s the real premise with these localized computes that we’re doing, to be able to focus on how do we make sure not just keep this data countrywide or regionwide where it is today, but how do we have this compute maybe statewide, or even citywide, and bring it as close as possible to you. So that way there’s a breach, it isn’t a breach of millions upon millions of people, but a breach of maybe 10 or 100, which really changes the threat pattern of someone trying to steal the data, or someone trying to use that for marketing purposes or whatnot.
Definitely a different dynamic if you will in this whole thing, that will change the dynamic.
Well this has been very interesting, and I think very informative for our audience. We usually like to end these things off Pele, to ask if you have any books that you recommend or resources on the subject, or anything related to technology computing etc.?
For me, my personal best friend has been Google News, and being able to just whatever I look at which includes things like fierce wireless and light reading, many other things like Bloomberg and being able to try and get all the relevant things. So, all that information that I gather for myself I make sure I transport on our own blog actually, blog.mutable.io. We do a weekly round-up going through all these articles on drones, on 5G, on what’s going on in gaming and IoT, and we basically make sure we have that. So, for me it really about just ingesting the latest, and then filtering out what’s now and what’s in the future, that’s what I do for the most part.
I guess like any entrepreneur operating this space, you’re full-on in terms of you’re focused on the business as well as what’s happening in the world that impacts it.
Well, thank you very much, this has been a great conversation and appreciate the insights.
Absolutely, thanks for your time Leif.