Everything EOS EOS Node Operator Round Table, 10 May 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Node Operator Roundtable - May 10th, 2023

Description

Dive into the world of EOS EVM in this in-depth discussion with blockchain experts. Uncover the mechanisms of node operations, proxy accounts, and gas fee calculations. We also tackle the challenges and opportunities of EOS EVM, the role of user experience, and the impact of gas fees on the EOS ecosystem.

00:00:00 Start
00:00:18 Leap software update
00:02:00 Incentives for running EVM nodes
00:15:12 Reliability vs Decentralization
00:21:16 Laying out a vision for EOS EVM
00:34:57 EVM mining
00:57:08 End

The EOS Network Foundation coordinates financial and non-financial support to encourage the growth and development of the EOS Network. We’re harnessing the power of decentralization to chart a coordinated future for the EOS Network as a force for positive global change.

// EOS Network Foundation //
https://www.eosnetwork.com
https://twitter.com/EosNFoundation
https://www.linkedin.com/company/eos-network-foundation
https://t.me/EOSNetworkFoundation
https://discord.gg/eos-network
info@eosnetwork.com
#EOS #Web3

A

Oh welcome everybody, May 10th back for another round table um and we will start with uh leap. Software update Stephen is out today so we'll hand it over to Brian to give us the update on the software side of things. Excellent.

B

Yeah, so we have a uh a couple of uh more firm things and one one looser sort of update, so NT uh V, 4.0.0 rc2, um we'll likely be landing early next week. That's what we're currently targeting for that um June. The 1.2.0 also will likely be landing early next week um and then leap itself. We have a patch release um planned, so there's been there's been a few.

B

um A few defects discovered that we're uh fixing. Currently um we don't have a date for that patch. Yet, um but a student as soon as possible, that'll go out so I'll I'll keep folks abreast in the um channels.

B

And that's it as far as updates go.

B

With that, um so we'll we'll likely end up heading back to the um the node types discussion. um I also have some topics here that we've sort of generated from a previous conversation that I want to.

B

um You know review at a top level just to see if any of those are are still uh relevant and if we should um discuss them uh further than we have to date.

B

um But before we get into that, you know, I did want to talk briefly about running. You know: evm EOC, VM, nodes and sort of like what Explorations or participation people have um have sort of looked into there.

B

um And what would uh you know what sort? What sorts of things would would help in that area, um so yeah I'm gonna leave it open ended? uh Has anybody on the call yet started um running be acbm nodes.

C

Was there a request to.

B

Run the ebm notes nope not today.

C

Just thought I'd start with that. One.

D

I had somebody ask me about it, but no we haven't tackled it yet I think we took a look at it.

E

B

And um so let's say that we I.

C

Know Stan set it up because he commented in the channel, but he set it up and he opened a bunch of GitHub issues with document yeah challenges he ran into, but yep.

B

So if we were to put out and say like hey, ideally, you know, ideally every nude operator is running this stuff. um You know what what might you, what might make you likely to actually do that? What might stand in your way? You know those sorts of things.

D

My first answer would be good luck to get everybody to do it.

C

Well, I was I would say the same thing because how many of okay, so this is eos right, so uh how many of the existing EOS node operators are running public I, assume you mean public evm? No, it's not private ones, public! Anything right so I mean the the number of people in that category is uh somewhat Limited.

E

F

uh Yeah I just arrived, I have no idea what you're talking about you showed.

C

Up late arrive, late, you're in church did I win a prize absolutely and buy more Hardware at your own expense, run the evm infrastructure for everybody on the whole planet.

F

C

Ovh will scale it's not. A problem is.

F

That a little Docker image, isn't it yeah yeah.

C

But but seriously uh Brian, what's the incentive to run the infrastructure right so in in the early days of the EOS slang that was like well, BP's, you know need to participate in the ecosystem and Mobility Community or whatever, and that was the incentive uh that ship kind of sailed. So now, what's the incentives.

B

Right well, I'll turn that I'll turn that back around what sort of incentive would need to exist uh to make it worthwhile.

D

Something to pay the bills, no.

C

Yeah I think that's the minimum right.

D

uh I'm, not even gonna, say it's a minimum for us USA, but hey, that's! Okay, nice.

D

um I mean a project: that's actively for us, it'd be just a project, that's actively engaged and wants to use it. You know what I mean around whatever the hell, if you're willing to give it a shot and use cases, but but let's give it.

C

A shot not like make it available for anybody- oh maybe maybe you are forgotten for punishment and you know having.

D

Yeah, probably a little more likely for that than than many others experimental.

C

When you say, I already need a service that checks these public apis. Like you know what use nation was doing with the validation stuff like you're, going to need some service. That's going to check that these nodes are actually running and they're, not problematic and then potentially work out, something where they get some of the gas fees.

D

I think that's the way the EV um node is bills right, I think it automatically has some sort of operator collecting some gas I haven't looked into it I think I, remember Stan posting about it, but I I think that that's some mechanism to it, but yeah I mean nobody's going to be using it. It's probably going to require you know a decent amount of hardware and yeah I'm sure it'll get abused.

D

um What were you asking you're asking something.

B

Yeah, so you you said: um that's for you, you would need a project that is actively engaged and wants to use. It I wasn't sure what you meant by a project, um our.

D

Threshold for operation is a little different than a lot of others, because you know a lot of our Hardware is already capitalized and and I mean kind of what we did, but we aren't just an absolute button for punishment as much to be saying so you know we're. We spent out Technologies and kind of helped people get going to even get started before they can afford whatever it actually is. You know a rack of I-9s to do it, but you know it.

D

We hopefully helped jump sort them as opposed to happen, to have some immediate funding. Now, if the network itself could find some funding, that's optimal, but that's kind of Shaq's point is then you get into the game of well. Is it really working? How much is it used? How reliable is it this that and the other right.

B

um But what I'm still not clear on is when you said project, do you mean an internal use case, or do you mean like a.

D

Client any use case somebody says: I need I, want to use an API and you got, and somebody did, one person did reach out and say: hey I want to know this API.

D

Ironically, they just wanted to shortcut into the peer-to-peer Network for their transactions anyway long story short I mean there are some projects that are investigating it, but you know just a use case that we can. You know all right sure we'll can throw ten thousand dollars worth of servers at it. Why not.

F

Well, I suppose we are what producers we get paid programmatically while the by the actual chain itself, um if this would uh improve the value of the EOS token and make sense for us to run it right.

F

Yeah- and you know you kind of support yourself to you- know, get get more use on the actual EOS Network and um you know the value of the token up, etc, etc. So, let's.

D

Start into it, yeah I mean yeah somebody to give some value out of it doesn't even have to be purely Financial. You.

F

Know uh well, for us we're running a business. We want to make sure that there's value and that value leads to growth of the ecosystem.

F

So yeah, that's it.

C

Where is this whole road map road map, maybe or what I mean just like for people not familiar with? What is this infrastructure that is needed to run for the eocpm? Are you talking about, like just you know, API endpoints to use with like metamask stuff? Are we talking about like apis that is like reading within the usuvm contract? Can, like you know, exposing some common uh uh things? People want like what is it exactly yeah and maybe people will be able to you know because they're just running another, you know small kind of service.

C

uh When you know the evm is not really used that much most people won't mind just adding that to one of their servers. But you know it really depends on the scope.

B

Right so there's there's kind of two sides to it. One is one is uh J, the RPC piece um and the requirements for that um as as it's currently implemented is that you need to have an in-sync ship node um and it, and it also has some sort of heavy lifting or like indexing and whatnot, that.

D

B

There um and then the other side, which I admittedly don't know a whole lot about, is uh at this stage. I'm still learning um is the minor in quotes, because um it's not mining like other ebms, but basically that that pushes transactions into the um into the evm.

C

Which is in the RPC endpoint the URL that the user is going to set within their metamask.

B

Yes, yeah yeah, um so.

C

Is it going to be like you're, providing like load balance, you know like? Basically, there will be like a little balanced, you know end point, and that goes out to the different providers or how are you even actually running one? How are you going to get it people to actually use it because people are gonna, you know, add the network only once most use it.

B

Yeah yeah, that's a that's a fair point so like at a higher level like how do you, even because, because in something we've met a mask, the RPC endpoint is sort of synonymous with the network. I said yeah.

C

um Randomly bouncer would work like some type of you know, round robin or least connected or something, and that will alsoever is running, that little balancer will be able to figure out who is getting what uh calls as well as who's responding correctly and if one fails move to the next one, and those statistics could be used to drive, who should get something out of say, a certain pool of gas fees or something.

D

That's been discussed on many of the different networks because of the same type, I mean even just standard node OS apis is to have a network common address which is intelligently, load, balanced and managed across in and out BP Network, and you know validated that it's functional, you know if it lag behind, you don't want people and, and it does happen and it causes people weird things. It'll get every other response, go get a weird response, or something like that and using that towards trying to know who's, doing the work and using it for billing.

D

But as you can imagine, that becomes a pretty massive product and and service, because it's so critical and becomes the central point of you know, attack mechanisms and, or somebody centrally asked to manage it in a decentralized.

C

Yeah, and on top of that, if someone is like say their ship, history is behind or something right and it's done at their turn. You know maximize makes like a billion calls like a I wrote. One like you know, RPC for kind of like antelope for it, and it makes like so many calls all the time to update all the assets for the different.

C

You know accounts that you have so if, for example, some of these go to a different loan balance server and the ship node is behind you gotta see your assets just value, just changing randomly and like it needs to like whatever, if a proxy is to be set or a load balance. So it's not that a video just like engines and boom, uh you know round robin and you're done. It's gonna have to actually find a way to check the current head.

C

uh Verify that whoever whatever API is gonna proxy to is actually in sync- and you know, can answer correctly.

D

All over the world things like that, you know I mean just it becomes this massive product of the network right, but this would require like.

C

Good health checks, like the the stuff that you guys are describing where it's caught up and ship is responding and all that stuff like we should build that into a good health check. And then the load balancer would only serve traffic to people who are passing the health checks, which is a problem that we're still working on. To be honest. But.

D

We are using load balancers for this and we have health checks so right.

B

Well, also, I don't want to continue on on one solution like load balancing uh to different nude operators, because it might be that that's not that's, not the right solution. I, don't know right. So my.

A

Question is: do we want to optimize for reliability or decentralization, or some mix of both like one solution might be? There's one operator that has a contract that has a highly reliable RPC node and it's just a direct contract with the DNS to you know your own type of service.

B

This is a this is a Brian statement, not an enf statement, um but I would say like based on what I know about uh you know how EOS is seen right. I I think we we're we're better leaning into like uh Speed and Performance than decentralization as the selling point right um and and sorry and reliability, I meant to say, um and so I that's my gut.

B

But that said, and that's why I said I, don't know that um load balancing is necessarily the right solution, because I, don't I, don't think I, don't think we necessarily have to um only do one or the other right like people could provide alternative points.

C

You know sure, but then, like when people go to the you know, evm official, you know say, for example, the website and click add to metamask who's. The address and gonna be there, because you can only put one address and once it's there, people are not gonna change it like.

D

Look at anchor and gray Mass I mean that you know we all run nodes, they can run it, but everybody uses gray mask and that's what it yeah I'm.

C

Kind of like at least with gray Mass, you can sort of existing gray Mass. You know when you're using War for USGS on the front end, you can specify your own kind of API endpoint. So a lot of people are actually using different ones uh from the site which you could kind of do with metamask. You could prompt the user to add the network again, but everybody does that it's going to be like uh you know, they're gonna have a like a ton of networks into mechamask. It doesn't make sense.

D

I I would be interested to say if the enf kind of has a big initiative to try to get. You know the evm projects going on EOS, so maybe look at having a standard load balance. Whatever you know, simple, that key providers could you know sign up, for it does itself check as long as you aren't, you know, do a bad whatever works out and offer that to the community as the default address and be able to say. Look if you want to use your own whatever, but I mean from a marketing aspect.

D

I think you really need to try to make it as easy as possible to add it to metamask and not have to go through all the different addresses and.

D

Most people are okay with. If you have a need and have the budget and requirements, let us do our thing through our reliability and participate, hopefully get compensated for it. So I think you would find a decent participation worldwide to say: I really want some good API providers, I.

C

Really like the idea of like us coming together and like providing a single endpoint, I think that'd be really cool. Like that's so awesome that you guys suggested that um if we put a bunch of different endpoints behind the same load, balancer.

D

Because right now we have it in two regions in the world, but.

C

Like the block, producers are everywhere, so if we had it in all these different regions, it'd be so much.

D

Faster and and better for people- and you know if you guys wanted to publish like if the if there's.

C

An issue with the load balancer: you can connect directly to us this way like that'd, be cool too, but I don't know, I think this is a cool suggestion. I'd love to work on something like this.

D

If it's an ANF initiative, that's good enough for us, we can figure out how to pay for it later and.

C

That's our threshold and I think it's the only solution. Unfortunately, the load balancing for the end user experience.

D

Correct the comment Daniel's Point suggested address.

B

I'm I'm not suggesting no form of load balancing, but.

D

B

You know I can't imagine uh different users having different um sort I'm looking for priorities and I'm, not saying this is a thing we should do, but like I could imagine you know. um Okay, do you want maximum reliability or maximum decentralization right and and therefore you're going to have two different sort of options? For you know ones like just push this button and it'll add metamask and the other one's like. Okay. Here are a bunch of uh RPC endpoint options. You know, what do you want to do? You know like have.

D

You ever added a customer who would ask I mean yeah.

B

I have it, but it's not that bad, but it's yeah I agree that it's not it's certainly not massive.

C

Down it's fine but editing it. Every time is not fine, but what you would do is basically, you would publish a page where the user is added to metamask, where you say here are all the providers, and that is the decentralized provider, so say all the BPS that are willing to do that and then on the top, the one the only one that has, for example, at metamask.

C

This is the approxy for all these different providers, so they realize that decentralization of it and you can have say multiple endpoints on different domains, for example like that are proxies all also, and then the users realize this. It's like in Fiora. You know like from an end point of the user. They don't care about decentralization, they all connect to.

C

You know inferior and ethereum, and and that's it, but then that's their source of Truth for assets and level even deeper, because it is decentralized in the sense that behind this proxy, yes, that you know domain is controlled by one person, but behind the proxy is actually leading to very decent services. That are all being checked for to be consistent. Basically, on a Ted and so on and so forth,.

E

A

All right, I'll.

C

Go back to comment I made earlier, which aligned with kind of what Ross is in. What's the road map, what is the need for all these zvm RPC endpoints right? So when we started with EOS, there was like this white paper that said: okay, here's what we want to do with the network? Okay! Well, it all fell through. We all know the story, but at least there was a vision right here. Oh there is yeah CBM okay. This is.

E

Their white papers or.

C

Blue paper, where is the park, then we can see, oh for XYZ, to happen. It means we need node operators to run rpcs. This is how they're going to get paid. This is the Boost right and then it's all laid out, and then we can see how we might or might not want to play a role in that vision, yeah yeah. Maybe it's been laid out somewhere that I'm not paying attention to which is totally possible uh because I'm not.

B

We need we need the um what's the word I'm, looking for the momentum that comes with sort of uh when you have a movement, so to speak right where.

E

People are bought into a shared Vision, so yeah.

B

Okay on the on the road map, I'm, assuming your reasoning for that is exactly that: I like to create that sort of uh shared vision and uh or is there some other reason that you're interested.

D

In what do you want me to do, and why do you want me to do it? Yeah.

C

Well, yeah that but more practically there's a hundred evm Shades up there, uh just because EOS has an ebm. Why does that put them more important than I? Don't know Avalanche or polygon or whatever right so or ethereum or ethereum, or you know, insert the one you want, but I like I, got a message yesterday anyway, starting up a new email saying you want to join all right. Okay, well, incentivize me to want to do something on this chain other than you know, increase the price of EOS, which within help us as plot producers,.

D

C

Right so it's like almost like you want to see it attracting, like uh you know, projects that that projects are moving from, say ethereum because they eat whatever more performance, slower fees or whatever the reason may be, and uh and you kind of have a place where you can track that this is actually bringing in users or see the benefits kind of thing. I'm just trying to understand your point Matthew!

C

Well, it's not that we have this I I, don't understand what the vision is. Yet it has. It been laid out right, so I I have a million things to do in my day. Why do I want to prioritize setting up an EOS dbm versus setting up a XYZ chain? Evm I think the whole point at least how I see it for evm is not to actually look we're an ebm chain.

C

It's more to just you know, get uh Outsiders, mainly you know from ethereum or other ebm chains to come and try to use it here at cheaper costs and and more performance and then benefit to eventually convert to to proper EO smart contracts. I thought it's more like a funnel yeah. Is that the vision I'm not sure, that's the vision.

C

Right, it is the goal to convert everybody. Dear everybody, dear world, come here get your thing running on evm on EOS, oh wait! Now. Actually it's convert everything again and run everything on the AWS. I, don't know I! Don't it's not convert everything again, like the whole point. Is they can take that existing code and run it simply on EOS and again I?

C

Get it right and gain advantages already, where some of these projects may be running into a lot of limitations on their you know, whatever natively they developed and then once they slowly, because it's not going to be a magic wand. Like hey convert, you know once they slowly see these benefits and then get more entrenched in the ecosystem. They can slowly start migrating their most, you know, say expensive tasks to another smart contract or and then eventually migrate, their whole thing.

C

So it's more like uh you're, providing a ramp up kind of for these external users that wouldn't otherwise come and learn EOS. Now they have a way to actually get on it gain some of the benefits right away without turning it ins and out and then have a pass to even expand on that much more in the future. But that's just my view, I don't know if that's, but what's a vision or something also enf for you know. Well, that's the thing I don't know if that's the vision or not.

C

That's my personal Vision, like speaking as an individual and not the enf like that's how I see it working I I identify 100 with what Shaq said so I I'm I'm playing a little bit of debit, double tap kit here in case you didn't guess right so I'm, just destroying swords or whatever onto the. But you know, I want to see this thing succeed. I would love to have a reason to run. It tell me what it is.

B

Yeah for that matter, it's so first of all appreciate uh the candid discussion, because that's I think where the best value comes from and um if, if you notice me being silent, it's because I'm actually getting a lot of value out of hearing. uh You folks talk to each other about it.

A

D

C

So over the last year, I've played around with a bunch of dbm chains, so you know I spent the last year doing that, rather than you know, EOS more specific things so which which one do you like the best and why I I'm not sure I, like any of them, the best, because I so I'm only looking at it from a node operation, point of view, I'm not looking at what happens on chain, I care, how easy it is to operate.

C

How does the consensus work? How does the software work? How is it backed up? How does it restore? How does this thing scale? Does it work? Does it always crash? Does it work with the latest Eragon? No, the thing crashes. Oh, you have another custom stock. I need to run just for your evmg great another thing to learn. Thank you right. So all these ebm chains popping up they all well I'll.

C

Do it this other way, Ray another thing to learn right, so the EOS evm, while I'm at the Miller okay, now I, have to learn the silk one thing and figure out how to run that great thanks right. So that's what I'm looking at so I don't have a fade. The the favorite one that I would have is the one that's the most boring.

C

That requires no work so question to you there would you appreciate it if we provided something like Helm charts or even just Docker containers that you could deploy without having to worry about what they are or is? Is that not something you're interested in, because I know a lot of you guys run bare metal?

C

Well, the not speaking specifically for myself, but there's three types of users that are node operators that I run across in my year of working on the stuff. There's the run everything in bare metal and don't give me any of this Docker crap people there's the uh I have a simple Docker setup. Please just give me a Docker compose thing and there's the I have everything in kubernetes, Just click a button and I, don't care just make it work right and the the amount of attention people spend on their nodes.

C

You know decreases between those three options. I would say right, so the I want to run everything on bare metal. Those people really care how it works or optimizing it and making sure it's performant. If it's click a button and deploy it in kubernetes, then well they're not really paying attention to what happening. They're just set it to latest and let it automatically update. Then you don't have to do anything.

C

uh I I tend to be in the first bucket, but I want the first bucket to kind of work like the third bucket right. I. Think mostly peas are in the first bucket and uh in 100 groups of Docker kubernetes are more or less like the same in a way because at the end of the kubernetes is a container system like docker.

C

That's all yeah, exactly and I think you guys use uh like uh not Docker but use a containerd right. We're using L1, bear mailers and ourselves yeah. Okay, all right so I think, like Matthew said like that, the important parts is not necessary that it has a Docker image that would help some people run. It I I, assume for sure for sure somebody who's gonna benefit, but something that could be run also on bare metal, simply update.

C

It easily has like clear tools to backup restore on error on Crash, like something that's like Stables to run without too much messing around is I. Think you know. So what are the parts of running the node infrastructure that I that I find? Is you have to get a feeling for how it works right? So it's not just set it up and let it run. You have to understand. Okay when it's overloaded. This is how it behaves when you do a consensus upgrades. This is how it behaves. These long messages are yeah.

C

You just ignore them. Who cares, but these ones tend to be important, are what's the alert levels? When does it get behind? You know how fast the disk storage May grow. Exactly can I run it on him yeah exactly so. All of these kind of things has did you don't typically audit. You can't Auto automate all that right, so you have to just learn how it works and you get a feeling and you play with it.

C

Then after a few months, then you're gonna, like okay, now I, understand how this thing works and then you're like happy and hopefully there's a you know an upgrade process that follows always the same way exactly so then you just have a recipe in your playbooks can just run and uh deploy the thing automatically the same way, all the time right, but there's a learning that goes to is it requirements and growth is.

C

It is just like very helpful like this, like the main question I would have is like how much disk space does the indexer that indexes I assume is some type of you know, database and all the stores? How much is it as current? You know in its current form, how much it grows over time. This is always the first questions, a no right. What what how much sleep should you eat my name?

C

How much this based on you right, because if you're running on hetzner, you have to buy at the right or you know, rent the right machine uh do I need an ax101 for this, oh or no one of these other ones right and it costs money we might have to change and yeah in the ramp yeah.

C

So having those uh you know, give me, the higher requirements for the next year uh is definitely helpful, which of course regards some crystal ball, because you know who knows what a year it was from now, but yep.

B

So, just to try to restate what I just heard one of the barriers is a lack of understanding of what the hardware requirements would be right like how much CPU do I need how much disk space and, what's the current, projected growth rates for those well.

C

Okay, so so cool are you depending on which kind of operator you want to attract? You need to give more details or less depending on you know who you're trying to do so. If you give us a million dollars a year and tell us to run a node I won't ask you any of those questions, because you gave us some money. I'll cover whatever resources, you need: okay, I'm, exaggerating of course, but if you're a dear little guy, please run my thing and you know you're getting fifty dollars in gasps a month.

C

Well, they need to be really clear about what is the requirements to run that node, because is it worth the 50 or not right so where's? The return.

C

C

Now I'm gonna be like apis like we're, we're just going to team people, we're not who are not running it like. That was the method back then seeing people and hope the voters do something about it and we saw how that went. Yeah. Well, that's not gonna work right, we're not incentivizing it. Anybody is introleted or reliable, yeah right exactly so. I'm like Michael I'll run infrastructure, even if it doesn't make a profit.

C

If there's a vision that you know it's helping people it's going to grow the ecosystem, there's a bit yeah sure we don't have to make a profit on it, but it can't cost me right. My time is the most valuable thing to pay me yeah. If you pay me, then the you know the requirement for believing in the vision goes down right. The more you pay me yeah.

B

On on the mining side, and so I know, most folks here uh um operate often as block producers on the mining side. If we assume that there's just ignore any diminishing returns on number of miners who are pushing transactions and whatnot, um so we assume that the most possible is is the best right.

B

um What and this might be speculative, let me know if it is, but what do you think the receptiveness of folks who are kind of in standby and lower positions, uh as not node operators um to running those sorts of things might be? And if you don't know you have any suggestions on how I might find out.

C

Well, first, we need to know when the what mining gets. Is this like a crucial Fork mining, where we can accelerate it to the GPU, where you know we can use an Asic where, like what algorithm you know like that, would really attack. Basically, a key processor.

D

Isn't it I mean this? Is.

B

Michael I didn't hear you I'm. Sorry, oh.

D

I as I understood it I mean basically it's just sitting there waiting and receiving transactions and smashing them into the chain kind of like a Oracle processing. That's like.

C

A transaction to relay node and they're not actually doing any block cropping like any block, Mining and stuff, because right now it's on the ER, so there's a little block mining, no consensus layer. So it's really just a transaction relay the fastest transaction relay. But ideally, if the, if there was a load balancer that would proxy you know based on different requests, then whoever was essentially load.

C

Balanced, proxied targeted would get the sort of that first response, but it can be another BP that could probably pick it up as well, depending on how the load balancer works. But uh from from my understanding it's more of a transaction relay node than a minor adjusting work.

A

So what it makes us have a scenario where it's like the RPC endpoints is a monolithic, inferior red type solution and the RPC and the miners are more decentralized.

C

Well, the the monolithic is the load.

A

Balance, the RPC is the big one right. The load balancer.

C

Is the central point- and this is I- think I'll I'm sure inferior is like that too, like they probably have multiple different providers within their infrastructure, but they have one endpoint and that one endpoint does the sort of like load balance, and the graph is like that.

A

But with the RPC and the miners, both we have behind the scene, the load balancer or at least two different things completely. Like would is there a scenario where you had RPC and wanted to add a low ad balancer or minors behind a load balancer with a decentralized set of room operators and an RBC endpoint? That's just one provider, although.

C

The providers would have their own little balancers, so, let's say like Matthew would have Michael would have their own load balancer within their own infrastructure.

A

Connected to a higher load balancer, which kind of does that Global redistribution.

C

And it's only because the users can only put one one address and that one addressed is you know it's it's whatever that one address user puts will be the first point of contact and.

A

That's for both the RBC endpoints and for the miners. This is not again, it's not two different accuracies. You.

C

Could go directly to the RPC endpoint directly or you can go through the um the sort of the global proxy and I think that was either Shack or uh if it was act. I mentioned the you would list all the different providers you can go and connect directly to them or you can connect to that higher level load balancer, which is essentially all of them, combined with the round robin the priority uh here.

C

You're still talking about that providers, proxy I think Daniel is asking about, like the miners, like I'm sitting up to what the miners actually do, because now it's saying it's kind of like adding like an API, that's just relaying stuff to the node, but it's releasing it from where what Authority does it need to actually write this data to the blockchain like who? How does this Authority change or the schedule of these miners, like it's been very unclear to.

B

Be honest to be clear that minor, you know, put it in quotes. I use that term, simply because it's doing it's achieving the goal that miners on other networks would achieve, which is getting transactions included right, um but it's not doing the activity that miners on other uh ABM um platforms.

C

So they're not competing is going kind of in a round robin between the different people in the miners or they're, not.

B

Another Frank thing is like I still have learning to do here. I, don't fully understand this part. Basically.

G

Yeah the way it works is that you know there's material transaction that comes in right and so that ethereum transaction needs to be uh wrapped into an Els transaction, and then the Eagles transaction needs to be submitted to the network right. So to do that, obviously someone have to pay the CTU and net with that transaction. So the minor is the one paying that CG and net for that us transaction that wraps the ethereum transaction into the network as payment for paying for the CPU and net.

G

Then the minor kit uh FC, which has been paid in the in the Els or in the ethereum transaction.

B

Right- and this is why there's a there's an advantage to having more minors is because it distributes the fees and and also the the rewards the gas fee rewards. So so.

C

It works similar to kind of I, don't know if you guys are familiar with the ether account by U.S Argentina, where you know you, you have this smart contract on chain, U.S account user account, and then you know using metamask.

C

You talk to an RPC, a custom RPC endpoint, that what is what it does is takes the uh and say that the transaction of a normal usio transaction and converts it into like an rlp type and pass it to the smart contract, charging, the user for the fees of CPU net and and even potentially buying gram and then sends back this fee at the end to whoever pushed it. In this case it will be the provider of the RPC, and this is similar to that from what it sounds like the RPC endpoint provider.

C

Is the miner yep? That's from my understanding. It's just that the transaction that essentially whatever pushes the transaction, would get the the fee collected for that transaction at.

A

A ratio because there's also Ram included as well so the smart contract itself, so the evm would.

C

Pay for the ram The Miner would only have the base for the net and GPU.

B

Yeah, so it's it's one, I'll call it service. If you will two separate jobs, one service right right. So it's just one end point, but.

C

Yeah, so maybe so I just separated as two things. That's really one thing.

A

Though it's one thing, so, if we did have it in Sierra solution, the inferior would be the only minor, they would be the providing an RPC endpoint and they would also be mining. Every single block there's no way to subdress those two. Oh.

C

We need to make sure then like the software, because the software is going to need to estimate the CPU and net before it puts it so usually, like I would use compute transaction to figure that out, but then compute transaction needs to point to an endpoint that is similar to what the producer is gonna process. So we see all these different minor slash, RPC endpoints will need to point to similar compute transaction apis or else every time it goes around. You're going to get different estimates for CPU and net I think that's a fixed.

G

Maybe okay, it's six for the interior! So there's it doesn't matter like you might want to do that in terms of like, if you wanted to did not mine, particular ones. I guess like if you were going to distinguish between one, that you were willing to mine and ones that you weren't going to mind depending on how much GPU and then there's something cost you but right I mean.

G

Obviously you could do something along those lines, but the idea is that you know it's: it's a fixed price that you're going to pay for mining that and then yeah. You know right. You know technically make different amounts depending on the actual CPU and that required for that, given turning back.

A

So it's a fixed. uh So basically you know when metamask is going to send and.

C

Each get estimate like call to the auto AC endpoint, so the RPC endpoint at this point will have the transaction in kind of metamask style or like East ethereum style and using that normally, what I would do is you know, convert it to this rlp stuff and do a compute. You know to figure out how much would actually cost, but here you're saying that just from that transaction I can have like a fixed formula to determine what I should return as the gas fee. Yeah for.

G

The transaction right in one hour alone, because it's just hard-coded return value, UK.

B

C

Buffer right, so it's you are spending less CPU than like you, the profitable yeah be profitable you're, not like in a negative. Every time you push a transaction right.

G

So the fee is enough to where no matter what the transaction is it it. You know you should be able to cover it with the CPU in that uh and still be able to pay for ram costs uh that are associated with. Although that's paid by the contract, you don't pay the right cost and then you know there's still a buffer, therefore, um actually making a profit on the call.

C

But if I'm running my iPhone endpoint I could choose my own gas sheet now.

C

Yes, it's a contract level, so that I think the gasket is a evm system contract. Well, of course, no the guess is Gonna Come From the node provider, no matter what it's been a cold and wood provider and give one of these estimate gags remember what is build.

D

I mean maybe that's more of a question yeah. What.

G

Build is going.

C

F

C

Exactly that's the point, I think you're, the user is just well man, that's more of a really deeper question. Maybe it's like you decide what to charge, because at the end of the day the metamask is going to make. Just this is estimate gas call to your RPC and you can return whatever value you want based on whatever you want, and that would be the fee that they see on metamask and that they will sign that they want to pay now. It's that c.

C

You know, like the smart concept, could do something where you know it determines the real fee and maybe returns that amount to the person or Charm. You know like it could find another way, but under this Paradigm, yes, like only the provider dictates when the C is because Master mask is only talking to that and that's what it's UI and the transaction it provides to you to sign will show yeah I think maybe the smart contract is better there's a minimum, maybe that's the only enforcement and then the rest of the percentage.

C

So if you you do charge higher or let's say the user opts into higher gas fee, they would just have to spend more. There was no actually priority and then there would just be a percentage, but the smart contract for the evm would have a minimum gassy.

B

Allocation I think that's what's in Forex. Yes, you.

C

Could have a higher fee and build.

G

C

um That's another check for you, Zach for your health plugin check what the estimate is and make sure the apis are not like you know, abusing their music, I'm I'm. Sorry I was actually talking in chat gbt. Can you please restate that again getting more information than uh it was about the Gatsby?

C

Maybe a check you wanna raise that it was about that the gas fee and the fee that metamask is gonna pop up and allow you to sign is only dependent on the provider, the RPC note provider because he's gonna answer that estimate gas call uh from metamask, so they can put whatever number they want. They can put like I'm going to charge, you know a thousand years or whatever it is, and that's what the user will sign and if the smart contract doesn't have restrictions, that's what will be pushed also to the smart contract. So.

A

C

Think, maybe in your health, you know endpoint at the stop that you need to check against to make sure that that Mecca mask or that RPC node is part of the proxy list. Is to check that the estimate that makes sense like a constant check, where you make sure that their estimate gas is not like you.

E

A

Not overcharged.

C

Should be within one standard deviation of the others or or something for example, but it's just a fixed formula. Then it should be. You know, kind of sick but yeah, like maybe some tolerance. Okay, thank you. I'll write it down.

G

It currently the the software- that's provided, grabs that each gas out of the contract um and provides it up but I, don't know offhand whether we then enforce the fact that um it has to match. Like I. You know we could definitely do that in contract. It's just not allow a gas theme unless it matches what we're expecting.

C

uh Take care of that issue at least yeah, because otherwise you know, if you leave it to the operators, then operators will do greater things because well they can change software yeah.

G

I'm not sure I'd have to dig into it and I don't remember force that or not, but we definitely could.

B

um So there are, you know, obviously some um questions and speculation about how things work. um Just testing so I know this has kind of been a middle area. So but what I've heard is um you know Clarity on how to participate? Clarity? On the you know, the vision and and trying to achieve a shared Community Vision?

B

um You know what I guess this goes into the Hal, but the the requirements, the you know. What's it actually going to take to to participate in this um and then, of course, a lot of questions about the specifics right.

C

um Yeah and I would also uh then the target audience. So if you want a ton of little providers uh like I want 2 000 validators, because that's a metric, okay, you want 2000 RPC providers. Well, that's a different requirement. Then I want you know the existing top 10 Tech EVPs to run right. So if we look at the early days of the EOS, EOS came with a Docker image right back in version one.

C

There was a doctor, it was provided and then at some point somebody made a decision that this Docker image allows too many people who don't know what they're doing to run EOS and it generates a pile of questions which we don't feel like. Actually, so, therefore, we will remove the docker right because that's what it did it made it easy to run the the ER software or the node. Yes right. So if you want that kind of behavior, then you need to deliver the software tooling to account for the audience that you want to run.

C

That's if that's my sorry long description and also one thing worth noting is generally an RPC endpoint is, is not using much in terms of resources. It just couldn't yeah, but you got. You got to run a whole ship node behind it. No that's what I'm getting at is. No, you like you need to have access. Maybe a ship, node. Okay, you need that I, don't need to necessarily run it.

C

So maybe the solution could also involve multiple properties where you have a proxy for ship nodes that are public, that this can connect to, because then you can separate the requirements because currently eat products running a ship. Node is heavy compared to anything else that you need for this RPC endpoint. That just needs to kind of you know, relay information and I'm not even sure why it needs ship to be honest because, like unless, like some of this data, cannot be gotten outside the history solution.

C

um Well, that ship could instead have that ship proxy we talked about, and then the proxy of BPS or of node operators have nothing to do with these people running ship infrastructure, because the the RPC code, when automatically connected to the ship rocks so.

D

I I agree with this public API if you're gonna run like evm and have swaps and stuff, like that, at least from the traffic that we see on our public API nodes, they're, just gonna, Smash and smash and smash, because they're trying to you know do that. So yes, the ship is the most intense, but now I mean to be honest with you dealing with the nasties and load balancing and having the public API presence, responding to whatever, in our opinion, is actually probably more intense than just running a ship node. The feeding, yeah.

C

But it's a slightly different skill set, but I I agree with you, so I could go. Have you can have both like you know the proxy for the VP for the normal API C? Can you know like have some type of uh you know? Chatting of the sea users have using bandham's rescue minutes, and you know, like those kind of like uh you, know, dynamic, firewall solution that can all be handled at that level where the end user is not managing that you know like the load. Balance itself is managing the list of bands.

C

You know uh uh like kind of limiting access to from abusers, and then you have the people running ship nodes. They also have a proxy that way like so many more people can run the RPC node because then it would be super light. There's nothing really in terms of database or state that you need to maintain and grow over time. Just need allocated, CPU and you'll be more. It has done yeah, so I guess two more things that came out of Shaq's comments.

C

Here is one in the graph ecosystem, so I'm going to diverge a bit. There is a momentum to split the who's, providing the infrastructure from who's, providing the graphs or the sum graphs. So you can look it up. There's a foreign proposal, it's under discussion, so the ship providers and the RPC providers two different things.

C

uh One person can do both the other thing is, you know, uh are we talking about building paid Solutions here right? So why does it all have to be free? Are we going to come up with a way? Maybe we should be having paid providers and working through it that way like the inferior for Eos Evian, which could be? You know somebody on this call yeah. It could even be a similar one where you get like free, for you know at least many calls a day, yeah exactly an average user and for abusers.

C

It has to you, know, use the code then headless limit, yeah and the way all those works. Is you get a unique API key which gives you a unique URL to that provider? So it's like inferior.com V1, slash, weird random numbers, slash whatever my RPC endpoint is or Json RPC. So there's no last thing after that, but anyway right, so everybody has a unique URL, which you generated from uh an API centralized back end yeah, exactly I, don't know, there's some options: there I'm a big fan of charging businesses money, but not individuals.

B

C

I think that's kind of that's. That's the idea right. If you're an individual or you make casual number of calls you're good, we don't mind, you can use our endpoint if you're sending us 50 000 a minute. Well, forget it hey.

D

B

Great, uh we I just noticed that it is 10 o'clock.

B

So uh I think this is a good uh point for stopping this um I have a few uh key takeaways um one is one: is that I'm gonna check in with the um uh product manager who owns the EDM product about um doing a guest spot here to sort of do a brief Tech talk on evm and do a q a and it could be valuable to um to have sort of the nude operator perspective, asking questions and getting answers um so um yeah and the other ones I've I've said a few times so I won't repeat myself, but uh thanks everybody for participating and for um uh Frank discussion.

B

uh Always always pleasure. Learn a lot from these calls and with that I'll turn it over to Dan.

E

Thank you very much Brian. Thank you. Everyone for discussion was uh great, dick and more into eocbn and uh looking forward to continue next week. Thank you. Everybody.

F

Cheers everyone bye.