Hashing It Out
Hashing It Out

Episode 112 · 1 month ago

Hardware infrastructure Pt. 1

ABOUT THIS EPISODE

Hashing It Out is dropping a series on the infrastructure of blockchain.

In this episode, Corey, Dee and Jessie talk to Rick Dudley , President of Vulcanize & TrueBlocks Founder, TJ Rush, about blockchain hardware.

Part one of this episode covers the infrastructure of blockchain hardware from the beginning to now. 

Part two will look ahead with DAppNode .

We didn't really introduce ourselves. Should we go? They know who I am, they know me. Welcome to Hashtan out, where we talk about web free blockchain infrastructure, the people who build it and why they build it. I'm your first host. Talk to Grety Petty. Now I'm Jesse Santiago. Welcome to hashing it out. This episode is all about blockchain hardware, how we got started and where we are now. De Ferguson here. We'll be talking to true blocks founder, D J Rush I love everything about the technology other than all the things that I talked about, and the president of vulcanized, Rick Dudley. I'M gonna go on a slight tangent here. This episode of Hashing it out is sponsored by Zen go. With a zen go crypto wallet, you can buy, trade and earn up to eight percent interest on crypto assets, while enjoying customer support and no more private keys. With Zen go, you'll get to mathematical secret shares, one stored on your mobile device, the other on the Zen go server, with no single point of failure. And something happens to one of your shares, your crypto is still safe. Download the Zen go APP and use the Code Trade Zen at sign up to get ten dollars back on your first purchase music. In this episode, composed by Nate Ferguson, we wanted to try and Suss out this change from the original narrative of Bitcoin, which was like one CPU, one vote and like people running a bunch of stuff at home, and how everyone participates in these networks, like commensurate to their ability to. I don't have one CPU or multiple CPUS, to where we are today, which is drastically different. What's your take on that, like how how do you see the evolution of of starting from this concept of one CPU, one vote in the in the hardware implications of that too, what it means to even run in for structure today? Yeah, I mean I think one CPU, one vote was a great um idea. It was a great aspiration, a great hope, a great um narrative, but I don't think it was ever really thought through all the way or very early on it was clear that that wasn't going to work and that there would be centralized Um parties that were basically like banks. I know there's somewhere in Bitcoin talk. Califinny talks about bitcoin banks Um. So I think that even though you could run bitcoin notes is for a while as an individual, either doing CPU mining or Gpu mining, and eventually FPGA mining was even Um, I think, achievable by the end user, like before the butterfly laps stuff. I don't I wasn't paying that close attention at the time so I don't know the exact dates. But after that I think the idea of individuals having reasonable mining operations kind of went out the window. And Yeah, why do I think that happened? Ultimately it was like, I...

...don't want to say it was bad design. I would say it was like more aspirational then, like it was more of an aspirational thing because I think the design. It's sort of obvious from the design. I mean the second or third comment in the original Cipher punk mailing list is you're gonna boil all the oceans with this thing. So people kind of knew right away that it didn't make all of sense. T J rush got the first computer. is when you discover the theory. Have that change over the years, right, because I remember when ethereum started, you ran missed and running a note on your laptop or Machigan or whatever. It was relatively straightforward. And simple. Um That's not the case today and it's certainly not the case. It's certainly differentiated from when we started in Bitcoin and running a full bitcoin note in your laptop was easy and you can mind. So like walk us through like you're path in trying to run this infrastructure locally and build local first applications and how that's changed throughout, like just your your experience in this ecosystem. So you know, the first the first thing that you had to do was run missed. You would download missed and run missed and it was called web three browser and inside of missed it started up the UM guth node basically. So it had its own a p I endpoint called local host, which served the RPC to miss. So that was in maybe march of, maybe February, March of. It took about, you know, a day or half a day to start a brand new installation of miss and then, you know, fast forward to about late and it's starting to take and there was this really huge d dos attack in October of and what happened is someone created about twenty million accounts over a ten, twenty day period and then they had to clean up that d dusk, which was another twenty million deletions of accounts. So what happened is the synchronization of the node went from kind of like this, it was like growing and then it went straight up, kind of. So it went from a day, a day and a half, two days, three days, and then everybody started complaining about keeping this note in sync with the network. And I remember Joe Luban and consensus in New York were writing all kinds of different pieces of software. Twenty thirty different projects going on and every one of them was struggling to get users because they'd have to tell them get the missed browser synchronized for five days and everybody's like, what the hell are you talking about? So Um Joe Luban solution was a reasonable one, which is he's going to stand up in Fura for every on it so that people...

...don't have to synchronize the note and they don't have to worry about that whole thing. And I remember thinking at the time, and I even said this on twitter. Probably you can find it somewhere, that's a really bad idea. That's a that's a bad idea because what you're doing is you're literally destroying exactly the thing that a decentralized network is, which is, you know, local node software running locally. And it's turned out exactly as I kind of thought it was gonna turn out, which is no one runs a note because not because they don't want to, not because they don't think it's interesting, because they don't have to. Why would they run a note? There's no reason whatsoever run a note. So Um now I actually I don't agree with that. I think there's a huge number of reasons to run a node, but the community doesn't seem to think that that's true anymore. Do you? Do you think that the hardware limitations are what make it so easy for a consumer or Usier to say, like, yeah, I don't have to do that? There's a whole bunch of reasons. That's certainly one of them. Is like the requirement on the hardware is owners it's it's really difficult. It's also you don't have to. It's like there's no impetus to do it. But for me what that does is it changes the nature of the of the possibilities that you have, because Infuria is shared among a hundred thousand people or or I don't even know how many people per minute or whatever. So no matter how much resources infuria has, they're incentivized to deliver as little bandwidth as possible to each individual user because they want to keep their costs down. So what happens is you get application software that can't really operate the way a fully local piece of software would operate if you had a local note. So when I run true blocks, which is my project, I'm running against a local node, I hit the Node Twenty Tho times a second, or however many times. I don't know exactly, but a hundred times more frequently than anyone can hit infura without getting rate limited. So the nature of the application that you can build when you have a local note is just a completely different thing than the nature of the application you build when you're hitting a shared resource like Impura. And that, to me, is what it means to be decentralized. It's decentralized. So the benefit of running a note locally to me is speed and not sharing and not being rate limited and not paying three fifty dollars a month if I want access to every piece of data on the chain, which I have to do if I use INFURA. I don't have to do if I'm running my own local note. But all of those things go away and people's attitude towards what kinds of things we can build, they just atrophy, they just disappear. People don't even imagine a possibility of building a different...

...kind of application. That, that to me, is the worst outcome of imperial and I don't mean to hit hit upon in Furia, it's just the nature of the world. It's like it just happened. Oh like, if you use Infuria as the the like canonical example of centralization. Right, there's this new friction that we're introducing with this decentralized network, which is running your own note. You gotta go buy a piece of hardware that maybe it's larger than your laptop. You gotta you gotta, you know, provision it, make sure it's set up correctly, you gotta make sure it's online, to make sure you have enough bandwidth, and those things have changed over the years in terms of the requirements to do so. So instead of that, in lieu of that, we'll look for you. That's right. And there's there's a trade off there, and you mentioned one of those trade offs just now, which is um you get rate limited. So it's within their benefit to Ben my costs to serve you just as much as they have to in order for things to work, and that limits creativity and building applications. That's exactly right. With a zen go crypto wallet, you can buy, trade and earn up to eight PC interest on crypto assets, while enjoying customer support and no more private keys. With Zen go you'll get too mathematical secret shares, one stored on your mobile device, the other on the Zen go server. Download the Zen go APP and use the Code Trade Zen at sign up to get ten dollars back on your first purchase. What did you use to for pulled storage? Um, there was a I don't know the technical details, I'm not terribly technical with the hardware, but it was a raid zero um thing, and I had I had actually two Linux boxes the same the same machines. So Um, this was open ethereum and at the time this was now about twenty twenty. So Um, open etherory was taking up twelve terabytes. Because I run my software wants to look at the history of every address. So I'm running an archive note and I'm enabling tracing. So it's actually the worst case scenario. For the disk size. It's not true of like a regular user who would just run the front of the chain. So I'm a totally different animal there. But then you start looking at Aragon and Aragon the note software called Aragon. This is all a theory made net by the way. I'm not too knowledgeable about other chains, but Aragon takes a twelve terabyte database and turns it into a one point one terabyte database and I can get exactly the same data from that, from that note software, as I could get from open etherory. And I'm first of all, I'm like, that is absolutely not possible to do, but he did it. So I don't know. I don't know what magic he did inside of his code, but he kind of set back the clock like probably five years on the on...

...this issue of how hard it is to run a note. So all the other nodes software, if I understand it right now, for an for an archive node, is still twelve or thirteen or fourteen terabytes, which, if you think about it from an infurias perspective, that's a disaster for them as well. You know, they can't be running separate notes. They're not putting fourteen terabytes of data on every one of their endpoints. There's no way. So what they've actually created it is some piece of software that stands in front of the node and you're not even actually getting access to the node. You're getting access to infurious version of the note which to me is literally google, literally Google. But so Aragon comes out and kind of saves the day, right. They push twelve terabytes back down to one and a half terabytes and it's now grown to about one, about two terabytes now. So I have I have like eight extra terror bytes, are ten extra terabytes on my machine. So I'm happily just running along for a couple of years without having to worry about it, I hope. But let me say another thing. Now comes the theorium too. I'm like, Oh my God, you know, now I have I can't even use etherium one clients without running etherorium two as well. So it just got twice as hard, not only for me but also for Infuria. It just got twice as hard if they're gonna run these notes. So I think what happens is people they I mean they have other priorities. Is the is the thing. They just don't have the priority that says run local notes. It's just not a priority and to me it's it's literally the only thing that we all need to do. And I'm probably crazy, but I don't know. What do you do? You guys think it's necessary for people to run notes, or is that kind of a radical idea? I run my own notes. I have a server. Yeah, yeah, or Corey does it do? Oh yeah, I run I run a bunch of different notes for different chains and different services. Yeah, I have a I have an since Aragon. I have an Ar Kind O. Aragon made it very simple to Um provision consumer hardware to run notes, and that was one of the things that like, I think it's difficult for people is not only like some people don't have access to these these this consumer hardware to run a note. And and most circumstances, even if you take like you know, the v M scales kind of terribly in terms of resource costs as it grows, bitcoin is really good in terms of scaling rescourse, about costs if you take out proof of work, but like the success of the network increases the computational costs in those cases of running the network. So people who get excited in the early days Um run things on whatever pard where they have...

...available to them. Like right now you can run at least with Nimbus. You can run ethereum two on on a Raspberry Pie. That is not advisable because eventually you're not going to be able to do that and see how to go out and buy new read buy new computers and so on and so forth. And so we have this situation that I find interesting, that I think it's very important that people run their own notes, but I'm also realistic in the fact that some people don't have access to these things. They're not easy to provision today. So like getting your nug buying the computers cool, getting it SINCD, like installing the software, getting the chain sync to making sure it's up, making sure it's up to date, getting notifications that you need to make an update, etcetera. That's all cognitive load and skills that some people don't have and that that makes it more difficult, or they don't have access to the hardware to keep up, or they buy it and then they can't actively grow it right because there's there's been circumstances where, like, I have a hard drive and I used it and then the chain fills it up and then I have to figure out how to migrate it and that's the way to start over. That's annoying, and so I think that's interesting. Right these computers that we have to run these notes, and it's very important if we want these networks to be different in any way from the Internet that they set out to kind of replace or transform or change. We need people to take responsibility running their own notes so we can build new types of software like the original Internet. But it's hard and I don't know just by experiencing it, from beginning of Bitcoin to today, and you look at the different available chains outside of Bitcoin, a ethereum, like our brand avalanche, like cosmos, Polka Dot, like all these different chains, they all have their own computational complexity and computational resources to run these things. And if you want to participate in a multi chain world, then you need to run all of them. It's like, are we expecting people have server rooms and like I'll gig a bit bandwidth, where Dudley says there is a solution to reduce the amount of hardware you need to run multiple chains. So I think I had an argument and when I worked at consensus in Um and we're talking about civil resistance and proofstake and all these things, and I think, I think that's what the conversation was about. And I was like, well, guys, money is centralizing. If you don't want a centralized system, don't allow people to buy the tokens. But that's actually the solution. Just don't literally, just don't sell them, like, distribute them some other way. Otherwise you're always going to have the centralling influence of money. There's billionaires, right, a billionaire has billions of dollars. There are people who have no dollars. They're...

...gonna have billions of times more power, Um, in the system. So so money is centralizing, right. So as long as people can buy hardware, Um, I think it's going to be. You know, money is going to be centralizing. Um, I think we can build systems. Um. So that's one comment. I guess the other comment is state growth in Ethereum is a is a mistake, right, it's a Um. The design was supposed to have checkpointing and truncation and we still argue about adding that into the network on a fairly regular basis. Um. So Um. So, yeah, so that's I mean, we can build blockchains today that use checkpoints, that have that have a fixed size of their state and then I forget that they keep they change the name of it, but I think it's called Mirror. Now. Um is an extreme version of this where their blockchain stays the exact same size. It's like some of the number of kill a bits, which I think is way too small. But you can make it any arbitrary size. You can make its three terabytes. I mean, you can make it whatever you wanted, Um, and we just and we have the technology to do that. We know how to do that. People just haven't implemented it. I mean we could implement that on ethereum like today. Basically it's not it's not super hard. So Um, people just choose not to, not to implement that, which is kind which is very annoying for me personally. Why is it annoying? Is that, like you don't like the concept of of like people don't like that because of the original narrative of being able to build from genesis with confidence and the what's the word I'm looking for? Um, counterparty risk of trusting other people for data, and I'd imagine those systems either or get rid of the data they no longer think they need so you can't do that. Or they use some form of cryptography to give you a different way of trusting the system. Yeah, I mean you would just use so it's the reason it's super annoying to me is because it doesn't fit into any data storage strategy that exists anywhere right, like, like librarians can't store data this way, banks don't store data this way, enterprises don't store data this way. No one stores an immediately available, monotonically increasing data set. It's it doesn't make any sense. It's not it's literally not operable. Like you can't run a system that works that way. We don't have infinite disks and and and you know, saying oh well, we're not, uh, you know, we're we're growing slower than the rate of disks Um, you know, disk size increase or you know, like Morris Law applied to disks or whatever, is like a bizarre cop Oup, like just build a system that allows people to use, you know, offline data, near...

...lined data, far aligned data, blah, Blah Blah, which is like a well established enterprise practice. Um. And so it's one of those things where it's like Oh, you people are fake, because you're not even addressing this basic issue of how do I actually operate this blockchain as an operator? So you mean it's tell me for a while now we would. We don't have to have like a super growing blockchain that grows forever and I have to like, you know, constantly update my hard drive space and like that's a thing that exists and we're just not doing. Yeah, I mean why would? Why would a client need to work that way? So I mean, yeah, there's no does the data set increase over time? The historic data set? Yes, it does. Does the Um, uh, the state object that ethereum has, for example, does that grow over time? That's harder to say because in fact it's very difficult to even meter or monitor that object. That's one of the things we do at my company. Is actually quite difficult to to know what the ethereum state actually, the state object actually is Um. But yeah, you could just say, okay, every year on Jan first there's a block number, we pick that block number, we generate a checkpoint from that block number and then a week later we published that that Hash and we published the state at that Hash and then we don't require the notes to have any more data from the past and people can just take that, put it on a tape drive, store on a you know, put it on archival tape, store on a tape. When they need to go get it, they can go into their closet, pull up the tape, put it toto tape and the tape drive and do their thing. We could just be doing that at whatever frequency we wanted to, every day, every year or whatever, and that, I mean, that's how real live production data sets actually worked. So then I guess it goes to what Coryer was saying, is that at that point you've got to do you trust third parties? Do you trust like or you don't even need to really, because if you're keeping your own archives, yeah, exactly, you just have your own I mean, think about think about it this way. I could just say we could just have said, okay, ethereum is never going to be bigger than five hundred gigs, right, which you can fit on your pinky finger, right, I mean five hundred gigs you can fit on an SD card, right, and you could say anytime we go over five hundred gigs, you know, we start we start this tape backup process and we take a snapshot and we build the tape back up and then we add the checkpoint into the new chain and you can truncate all that data. I've I've written, I mean I've we've developed this multiple times that have vocalized different different times in different situations. So you can just truncate the blockchain. With a zen go crypto wallet, you can buy, trade and earn up to eight percent interest on crypto assets, while enjoying seven customer support and no more private keys. With Zen go, you'll get to mathematical secret shares, one stored on...

...your mobile device, the other on the Zen go server. Download the Zen go APP and use the Code Trade Zen at sign up to get ten dollars back on your first purchase. So I'm running. I'm running a node for avalanche. In fact, I just spun it up like a couple of days ago. And when you're talking about the problems associated with running nodes in a home setting, it kind of like makes a lot of sense and I'm aware of this. I just know that a lot of people um are curious. They like, like I'm curious, to just play around with it. I'm wondering for blockchains to have real application for whatever somewhat idealistic but more practical applications for maybe not a global computer, achieving global consensus. Where is the practical implementation and what is the what is the hardware? That seems reasonable to be able to prescribe people to go out and buy and to learn to maintain and and just to to participate. Oh, you kind of mixed in a couple of different questions in there, I think you. I think you meant it as one question, but you at the end you sort of like split it into the last sentence, like split in the three. So like, Um, if you want to play around at Your House with avalanche, then just buy the hardware to do that. That's fine. Just realize that eventually, just naturally, you're going to have a very difficult time maintaining that note as your disc fills up, basically, and if you if you're not doing pool management, pool disc storage, you know, we use the FS pools, that vulcanized zone. So with the DFS pool you can do things to help, but ultimately, Um, yeah, and you could potentially, you know, run that for a very long time, like probably like twenty years or something and be fined, Um, in terms of what you actually need to run a smaller chain. I mean you need practically nothing. Um, you know, you could run it on the cheapest computer you can. Well, you know, like a raspberry pie or something like. You can easily have a whole blockchain that all the nodes are just raspberry pies and you have the same level of security that you have with bitcoin or ethereum. Um, that's that's possible. We don't, I think. I think. Well, the way that that would be possible would be too would be too Um. Manufacture a proprietary a sick and then distribute that proprietary a sick, uh in a judicious way, and it would still be affordable. Um. And then, basically, when people finally crack your a sick and start competing in the asic market, you either sue them, which is, you know, not cool, or whatever it it would work, or you release a new a Si...

Um, and that would work. I mean, depending on exactly what you're doing. That that could be sustainable, because it may be that you have so many notes, Um, and you're not giving out rewards, you're not giving out in crazy rewards that end up, you know, inflating in relation to the dollar. In this crazy way, um, and so no one bothers to crack Your Asi. I mean there's there's, you know, countless asis out there that I've never been reverse engineered. Right. So you could be running your you know, you could say, okay, we're gonna sell these a six, we're gonna sell a hundred thousand of them, Um, and you can order replacements from us for some period of time and they cost five bucks and we only give you one to one person and be mostly okay. Um, you can distribute them via lottery or something. I mean there's all sorts of ways you could address that in your have like as a happy meal toy. Seriously, I mean yeah, Um, you know in one of those claw things where you have to get the stuffed animal with the claw. You could put them in there. I mean you could distribute it any number of ways. It would be fun. Is it worth creating, like a system of people that are not economically incentivized to remove the Um the interaction of permission list and incentivization as a civil mechanism? Like, is there a way to build? I mean this is this is just nobody has this answer. Probably, but are there better primitives in terms of civil defense mechanisms that there's like a little bit of economic gain, but not entirely, but something that just makes the whole thing more fun. I guess. Well, you did it again so we could make a system that was that was fun and didn't really in the funness was sort of overwhelmed the economic benefit. So the systems also. It's funny. I mean, we've been talking for a while. I personally don't believe that global permission list systems make a lot of sense. I again, there's a very, very narrow use case. I think it's great for gun runners. I think it's great for I mean, I'm I'm a black person in America. You know, uh, there's plenty of times where people should break the law. Um, I think it's an ethical thing to do. Um, but you know, if you're not trying to break the law, UM, regardless of the ethics of it, the permissionless global transmission properties of Bitcoin are very weird. If I if I'm in a relationship, financial relationship, with someone, I know that Counterparty, I mean, I'm transacting with that counterparty. I don't need to be incentivized outside of that transaction. Right. If, if I'm if I'm trying to ship products internationally over the you know, across the ocean or whatever, and I need...

...to use this, uh, a legal payment rail to do that. Uh, I don't need to be incentivized to run the hardware anymore than the incentivization I have of running my existing business. So you don't you don't really need. You know, the incentive structures that we have now are these sort of pyramid schemes, because the chains themselves have no other utility. If I if I just wanted to use the chain to facilitate international business, which I think is a totally legitimate use case, a killer use case for blockchains, I would say, Hey, you know, shipping company, a, shippy company, B, Shipping Company. See, all you guys do deals together, all you guys work together, you guys have the same customers. Bring your customers together. And you know, you work in the Pacific, you work in the Atlantic. Like, you guys don't really compete. Let's build an work together. I'll give you all the hardware. Uh, you know, all the money that you have to pay in, like all the all the lost opportunity, all the opportunity costs, all the legal costs, all the friction. We're going to remove all of that by building our own little monetary system amongst the you know, twenty seven of us or whatever, and you guys give me, you know, a sliding scale percentage of that, so that I'm not, you know, out here, you know, in Dune Land, floating around on all my money, Um and and everything will be fine, right, and and and there's people who do build products like that, right, those products exist actually, and they're not talked about. Like I was surprised. I talked with one of the founders of skew chain and I I had heard about skew chain many years prior and I was like wait, you're still that? That project still around? We're like yeah, we do billions of dollars a year or whatever. I was like shocked. Right, nobody cares about that stuff. But that is a perfect, the legitimate small permission system that couldn't exist without blockchain technology. Is a financial application, has benefits to its users and doesn't have any pyramid or Ponzi scheme notions involved in it at all. Um and I think generally speaking, proof of steake systems have this property, and this is actually why I'm interested in proof of steak. And so when people say and and I and I believe that you should know your counterparties right. So I imagine a proof of stake system has, you know, under fifty participants in it. They're all named, they all know each other, they're all comparable in some sense. They can't. It can't be like there's one overlord guy and then there's a bunch of peons like that's obviously not gonna work. Like they have to have some sort of quid pro natural quid pro quo. But they are also trying to underwrite each other's transactions. Another example that I've worked with Um over the years was was uh royal t organizations for Um for music. So the people who buide sell...

...music rights and and get paid publishing royalties. Um, it's like a really sharky business and they and they have all this weird horse trading bs that goes on and it's really inefficient. And that's why, if you're into music, you'll see music like pop in and out of spotify. Like the reason like sometimes you'll be able to listen to us song on spotify and then like a week later you won't, is because spotify had the wrong royalty information in their system and there's no global way. It's like a huge best a blockchain would sort all that stuff out and it actually would solve a problem that they realized they had in the nineties and they didn't have a technical solution. Blockchain is a technical solution to that. So those are the types of so I think that, yeah, you need something like bitcoin Um and then you need a bunch of proof of stake systems like like millions, literally millions of them Um, and then they aggregate up and and eventually they have an ultimate settlement layer. And so practically I don't think that will be bitcoin. I think it will be ethereum. But Um. I think ethereum will be this global settlement layer for these Um for this broad base of millions and millions a proof of stick chains. Earlier more from Rick about the future of scale and economics and blockchain in part two of this episode, plus how dap note got its start and what the future looks like to co founder a tun and business and ecosystem lead pole Lansky. Part two drops August twenty one. In the meantime, you could find Rick Dudley on twitter at a F Dudley Z row. CHECK OUT T J rush at true blocks, dot io and depth. Note at depth. Note DOT IO. Curtains closed.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (118)