YTread Logo
YTread Logo

The End of Cloud Computing - Peter Levine

Nov 22, 2021
I'd like to show you a little bit of what I think the future will be. I will push them to the limit. No pun. I know what everyone is thinking. How can I say that?

cloud

computing

is coming to an end when it hasn't really started yet let me show you and this is what i think you all probably think i'm crazy if i stood here 20 years ago and told you that microsoft windows might be gone in the future or 30 years ago Digital Equipment Corporation might be out of business or 15 years ago Sun Microsystems might not be here you would probably put the same slide here and say boy has that guy gone crazy with these companies or these technologies they will exist forever everything that is popular in technology is always replaced by something else, it always disappears and that is the good thing, the opportunity and the beauty of the business, that these things really disappear and become part of our work as investors. it's looking at a It's not where the drive is today, but rather where the drive will go at some point in the future, usually five or ten years from now. a lot of what I think about is actually a very simple exercise that I encourage you all to do if you ever want to predict the future subtract something that is important today and fill it in with something else take something important take it out and fill it in with something and you start thinking outside the box out of the box instead of sequential as we might say oh there's more

cloud

computing

more of this I think if you subtract something you can actually fill it with other dimensions that you might not think about if you just go sequential and order something In his mind.
the end of cloud computing   peter levine
I did it with cloud computing about six months ago. I started to think. I call it my Forrest Gump investment rule. It's so simple it's almost stupid that you just add something and replace it with something else and about six months ago I started thinking well what happens when cloud computing goes away yes of course for myself well it's so popular in this moment, how could it disappear? I think it's really happening right under our noses now let me show you and let me explain why I think that's happening the world today is a centralized world and the processing that's done when we pick up our mobile devices the central cloud is we're all processing where the data feed is done when we do a google search or whatever you type something on your phone, that little stream of data goes back to the cloud, it's processed, the data then comes back to your phone, all of that is done centrally and to some degree your mobile devices are an endpoint or a display vehicle for what is happening in the cloud itself so we are now living in a very centralized model of the world and when i thought about the end of cloud computing i started Let's think, well, maybe our mobile devices will get a lot more sophisticated and we'll start thinking about when you, when we text each other. it's text, when I text someone in this room, um, my text doesn't go directly to you, it goes to some data center in Norway and then comes back here like it's the central hub and I spoke.
the end of cloud computing   peter levine

More Interesting Facts About,

the end of cloud computing peter levine...

I texted it that it goes to the core cloud, then it pushes it somewhere else and I started to think maybe it's mobile that will become this next generation and obviate the need for the cloud and intellectually that becomes It became a dead end and I started to think that it wasn't mobile but all the other things that are going to be at the edge that are really going to transform cloud computing and put an end to what we know cloud is and, so the change is that the edge is going to get a lot more sophisticated, not with mobile, but more broadly with the Internet of Things. a lot ab these robots drones and all the internet of things objects that will be created in the next 10 years think autonomous car it is effectively a data center on wheels and a drone is a data center on wings and a robot it's a data center the center with arms and legs and a boat it's the floating data center and blah you know it just goes on and on and on and these devices collect vast amounts of information and that information needs to be processed in real time that's the latency of the network and the amount of information from many of these systems there is no time for that information to go back to the central cloud to be processed in the same way that a Google search is processed in the cloud right now and this change it's going to obviate cloud computing as we know it this is a circuit board that's inside a car nowadays a luxury car not a self-driving car right now it's got about 100 cpu's today so what What we think about is self-driving a self-driving car in the not-too-distant future.
the end of cloud computing   peter levine
Ure can have a hundred of these cards, maybe 200 of these computers that are inside the card, so it becomes a data center. You know, hundreds of these connected computers in a car become a data center and then you think about connecting thousands of cars together, it becomes this. massive distributed computing system at the edge of the network and for those of you who are close by, we are entering the next world of distributed computing just like we've seen in the past, that's kind of a return: it's literally Back to the Future in what we're processing is done because of these very, very sophisticated endpoint devices, so it's interesting to note that in addition to these devices getting sophisticated, it's really about the data and for the first time if you think about these endpoint devices Endpoint For the first time in the history of computing, we're collecting real world data about our environment, whether it's vision, location, acceleration, temperature, gravity, information, but visually we're collecting the world around you. s through very sophisticated sensors so far computing has been fundamentally us humans typing things through a keyboard or a computer generating information from a database or generating log files or whatever but this is the first time that we began to collect information from the world. and that data is massive, the other part is real world information coupled with the idea that real time data processing will need to happen at the edge where the information is collected, you know, I use the self-driving car as an example. if we had a weight, let's say we're collecting real world information in a self-driving car, so you're collecting images, and as part of an image, there's a stop sign, well, if you had one, you know, take that data and send them to the cloud to decide there was a stop sign or a human being crossing the road, that car would have hit the stop sign and run over ten people before the cloud came back and said, "Hey, you should stop", the notion of real time becomes a very important ingredient given the massive amounts of information from the real world, we are not talking about text information, it is from the real world through the collection of videos and information streams, so both must work together and data will definitely drive this change.
the end of cloud computing   peter levine
You mentioned distributed computing, now we're going back to Back to the Future, think about the trends in computing, we started with mainframes as the beginning of the modern computer age, which was a centralized model, then we moved to a client-server model. back in the 80s and 90s and that was a distributed model the mobile cloud brought us back to a centralized model and believe it or not we are going back to a distributed edge intelligence computing model that is absolutely thematic with trends in computing going from centralized to distributed the other thing that happens, which is interesting in terms of the number of users that are actually a number of systems that exist. within each of these trends we have a massive uptake of users and devices that are out there mainframes at best there were 10 million people using mainframes with terminals and there were 10,000 mainframes thousand people per mainframe call it 10 million the Mac people on mainframes when we moved to PCs we had about 2 billion people using PCs so we see these orders going up now what's also interesting is if you think about the total addressable market I've always thought about a total addressable market for computing as the number of humans on the planet, after all, once every human has a mobile device, why do we need more computers?
Well the internet of things actually takes us to a whole new dimension because computers are no longer connected to humans so I can imagine where we have five billion seven billion mobile phones let's say everyone on the planet has one mobile phone call it seven billion i can imagine a world where there will be trillions of devices out there so if you think about the challenges of security data management all the issues of distributed computing now is the opportunity and the challenge of what we will all face in the not too distant future, is happening. right now it is starting with cars and drones and will proliferate to many other devices in the not too distant future.
The other ingredient that I also find interesting at this point is the intersection of machine learning and all of this endpoint data that we have. I am collecting. I think machine learning actually catalyzes the adoption of Edge. There are massive amounts of real world information that requires a machine learning approach to figure out the nuances of the real world, so the only way we can look at a picture or look at the massive amounts of data is with machine learning and that machine learning, algorithms and applications that employ machine learning will run on the endpoint, it's not going to be mach e-learning runs in the cloud, so what does the cloud become?
In this case, the cloud becomes a place where learning happens and a place where I'll call it the ultimate storage point. Processing in this new world will move to the edge, and that's where we'll make the most important decisions at the edge. This all matters because it turns out that humans are notoriously bad decision makers. I have a Tesla and I think autopilot which is like a very simple step into this world of automation is hugely beneficial as it's one of the most amazing it's a great innovation in terms of helping it's my autopilot , he's a much better driver than me even with all his flaws like me I'm a terrible drought because I write and I know I yell at my kids and whatever so autopilot doesn't have any of those issues just keep going the way, it's amazing, we can see that through the correct use of the data machine Knowing that these items are going to be much better and much more useful to us than we ourselves can be useful to our own data, so let me take you to the edge and talk about what's happening at the edge and then cloud centrally at the edge. there are three things that are happening sensing inference and action actually there is a very interesting parallel here with fighter pilots there is a framework for training pilots that was developed by Colonel John Boyd who was a top fighter and had invented this feedback loop which is called the youdo loop and Otis stands for observe, orient, decide and act and the bottom line with John Boyd was that if you could take a fighter pilot and create the fastest loop in your brain, your brain to process the information around this is in a dogfight that you would win every battle against the enemy because the processing that happens in my reaction to that loop to the extent that I could do it faster than anyone means that I went and the idea of ​​this loop OODA was prioritizing agility over power, and when we think about this new edge computing paradigm, it really is agility over power, the endpoint device isn't it's as powerful as the cloud itself but we can be much more agile because of that information loop it goes very fast at the edge processing only the information which is and what I think is over time the beauty of machine learning and what goes to happen is that cycle and I will call it the sense in an act. go faster and faster as processing gets more powerful as machine learning gets much better that cycle will get tighter and tighter against new information coming in there is a great parallel between this new world and how they have been really developed the frameworks for fighter pilots so that happens at the edge agility at the edge and then the cloud has a purpose the cloud will be about learning and that happens centrally so I'm going to take all of this information . the information that I have at the edge I'm going to connect these devices I'll curate that information and the learning will happen centrally and then I'll propagate that information to each endpoint device creating an ever tighter and more agile loop that happens at the edge, it's Ok so I'll talk about a couple of these things, the first is that the hubs will include cameras, depth sensors,radar accelerometers and they will be everywhere generating massive amounts of information as an example a couple of interesting examples on how much data is actually collected a self-driving car generates about 10 gigabytes of data per mile that's a lot of data a Lytro camera which is a data center in a camera generates 300 gigabytes of data per second just massive amounts of information and it's not just complex devices like a car or there will be camera sensors on almost everything I like use I put a picture of a running shoe here because in the not too distant future we are going to have sensors and running ning shoes that sensor will collect information your running shoe will run a machine learning algorithm with a very fast loop on it and when you are running you will he will say hey you are going up the hill you should shorten your stride or you are not doing as well as yesterday or you should pack up for today and go home and d get a little rest or whatever, but the shoe is actually going to be smart, so when I talk about these devices, it's not just big complex systems, but it can also be very simple things like a running shoe and, however, that will generate a lot of information based on the world around you. the data collected is extremely valuable, but it is also too much to be rejected every time.
Imagine if you were to return 10 gigabytes of data in a car for every mile back to the centralized cloud for processing it's not going to happen so it has to be done at the edge inference the data being collected is very unstructured it's first time seeing such large amounts of highly variable, unstructured data, so the inference will come by machine learning extracting relevance from the data itself whether it's a stop sign, a person, a tree, is running cadence, all that relevance will be extracted through machine learning algorithms this will give us powerful task specific recognition that will require training and data to keep that cycle ever tighter action is the final item after we feel infer now we have to do something In this regard, the critical security response capacity of perimeter systems will require timely data decisions. or real which is the latency between the edge and the cloud waiting for the response is going to be too slow but it was around when the world entered the first wave of distributed computing and it was exactly the same argument if I had a workstation on my desk and could process at the edge I don't have to wait for information to go back to centralized back in those days it was a server, data accumulates at the edge and processing stays close to the data and as IOT devices become more sophisticated and smart, the se Sensors and computing is going to increase so we're just seeing the beginning right now literally in the future there will be sensors on everything there will be machine learning algorithms and this is going to get more and more complex in terms of As much data is collected in the cloud itself, the cloud becomes a training center for all this information, one of the fundamental aspects of the Machine learning is that machine learning needs a lot of data to learn and so this new edge cloud model that I'm describing here has the beauty of a lot of data coming into a centralized repository to get smarter with that information. and so what's going to happen is we can have hundreds of thousands of cars collecting information, that information will be curated at the edge, so not all of it comes back important information from billions of devices coming back to a centralized cloud where the learning happens the cloud will store im important information and then the learning will propagate to all the edge devices and that's how things get smarter so the cloud has a purpose that doesn't completely disappear and by the way how the SAS applications and all that.
You will continue to work in the cloud, but this new model will really take it to a whole new level, here are some predictions for what will happen from this point forward. The data explosion from a sensor will kill the cloud. Sensors will produce massive amounts of data, existing infrastructure won't be able to handle the data volumes or rates and data will get stuck at the edge and computing will move along with that data at the edge we're absolutely going to come back to a point a peer-to-peer computing model in which edge devices connect to each other creating a network of endpoint devices similar to what we saw in the original type of distributed computing model, this will impact a wide variety of things including network and security if you think about the security challenges of all this data being collected and the network challenges of connecting trillions of peer-to-peer devices that are communicating with each other and probably processing information together without the knowledge from a central The information set is going to be one of the spirits with the opportunities or challenges of this next era of computing the other thing is we're going to move into a world of data centric programming it's kind of interesting in the same way yeah , I think the cloud is disaggregated in this new model, we're teaching everyone to code now we're teaching everyone to code the logic that is if-then-else and that's what everyone is learning to code well when dealing with data we don't use if-then-otherwise we'll use data to solve problems and the next generation of coders will be more mathematicians and data analysts instead of if-then-other error and therefore there is a transformation.
As for the kind of talent that we're going to need in terms of this particular information, I also think there will be new programming languages ​​developed specifically around the notion of data processing and analysis that are very specific to this kind of use. cases and eventually edge processing power goes up and price goes down we've seen it with every transformation of computing where we have these massive supply chains so if you think about mobile supply chain it's five thousand millions of mobile phones. a large supply chain is actually twenty billion that has been sold five billion in use so twenty billion of these devices over time created a very interesting supply chain that's why memory, CPUs, networks, and everything else in a mobile phone is so cheap because 20 billion of them have been produced, imagine one, now there are trillions of these IOT devices, think about the supply chain impact to help market the processing power and the sensors, as anecdotal evidence here, the current iPhone 7 has 3.3 billion with B transistors, the original Pentium processor, right, and one of those big sheet metal cabinets in 1993 had 3.1 million with transistors M, so think about the power increase. between then and now this is just going to accelerate that from a cost standpoint we're already seeing sensors mean all the magic here is sensors have to come down to a cost where I can put sensors and slippers to running so it can't be hundred thousand dollar sensors that go into running shoes; otherwise you won't buy a lidar for running shoes which is used for detection in self driving cars you may have seen that in addition I don't know if anyone saw it or I saw pictures of the car from google there is a facility lidar on top the first lidar for a google car was seventy five grand now why dar is under five hundred dollars and i bet this is like there ain't no self ving cars out there i bet that's going to 50 Cent's in the not too distant future like that's what happens so energy goes up by several orders of magnitude the price goes down what happens I can put sensors on everything literally in every shoe what what you want them with glasses and their ears wherever and everything that needs to be connected the whole world becomes IT domain so for those of you who think your job as CIO or IT manager has become easier, this is truly the opportunity of a lifetime they are moving from five billion devices to a trillion devices that all need to be managed all need to be coordinated together and all industries will be subjected it's not just like driverless cars if i'm in the insurance industry and I run a fleet of drones to inspect houses, who will manage it or I run an ITA Bay healthcare organization and they perform remote surgeries using robots and so many applications will emerge that will combine what we think of as a kind of consumer ori The applications developed with enterprise manageability they're absolutely going to happen, so once again, as we saw with distributed computing in the late '80s and early '90s and what What we've seen with the cloud and what we've seen with the mainframe, this is a huge disruption on the horizon that will impact networks. it will have an impact on the security of the store's computer programming languages ​​and of course management, so I encourage everyone to prepare for one of the biggest transformations to occur in the computing landscape.
It's happening right under our eyes and what do you think? now well thank you very much

If you have any copyright issue, please Contact