Filecoin Compute Over Data Working Group, 6 Dec 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Compute Over Data Working Group 9th Session (Phala Network)

Description

Phala Network: www.phala.network

A

And we're live all right, lovely, so everyone, that's watching the recording and everyone on the call. Thank you so much for joining computer working group I believe this is our night session. So we're very, very excited to keep the momentum rolling and on today's session we have the team from Fallon Network. We have Zoe and shelvin who are going to take us through the product. We've been building very interesting architecture, I love a lot of things about it and how it's native to their their blockchain Technologies um another different type of compute technology.

A

We also might learn some about the beta program that they're willing out and how we can get involved as well. So so, if you're ready, I'll go ahead and hand it over to you- and let us take us through your content,.

B

Yes, perfect, um hello. Everyone Zoe here from final Network um I, would like to ask Trevon if you could share his slides already. If you have access to it and um Trevin, do you have? Can you share slides already.

C

Yeah sure let me do that, but.

B

Because you just want to give your shot like um basic overview um on the main things that you also know who the team behind file network is so here today myself as VP of growth and Dr Chevron, our lead researcher and the team, but also wanted to mention Our Two, co-founders, Marvin and Han.

B

um Both of them had have like a quite a journey in their web 2 space, but then very early on, like very early in 2018, actually realized um the power of decentralization and we're very motivated to to build a decentralized cloud specifically for a computation part.

B

um They know each other and I think Dr Sharon, also for quite a while, and they overall got also very very fascinated by the vision that Gavin wood puts out while he actually developed polka dot. So fala network is the power chain on polka dot, as well as on kusama, um just to let you know which ecosystem we are working and the team is now plus 25 team members. So, just to give you this background and then um again, Dr Steven will go into very like the the details in the technical information.

B

I just want to give you um an overview of what kind of cloud we are building and how powerful it is. As of now so file network is really um the main photo of computation right. We don't um store any data. We work here with Avi for like five corn or ipfs together to actually um integrate through apis, but we are very heavily focused on the on the CPU computation, so it's a P2P, node, Network and I'm just wanted to give you some numbers here, so you can put it into um relation already.

B

How big our cloud is that we have built in a decentralized way over the last three years, so we have over almost 150 000 virtual CPU um globally distributed. So all of these servers are supported, of course, by the community, nothing owned by fala or the company behind it, and we also have a very high um education rate. So this is just some numbers to to give you this into perspective, and um with this I would already lead over to Dr Kevin.

C

Okay, thank you. Zoys.

C

Yeah, so actually I thought it has introduced. Actually we are a decentralized and the permissionless uh Cloud all right who served as the infrastructure for the web, 3 world and I. Think one of the very important motivation behind our work is that we have observed that the current decentralized apps is becoming more and more complex yeah.

C

So, for example, let's use the very popular game, the Stefan as a as an example, so like the modern uh decentralized uh apps, they do not, they should not only need to handle like the Unchained data, like the the nft, like the token part, so I think another very important things is that they need to handle some like the option: data like the, for example.

C

uh If you are like a Stefan user, then you first, you need to uh like buy some equipment like the shoes and actually the shoes is an nft, and it has a tribute you can equip it and to to gain more profits based on. If you have a very good shoes, but when you are really playing the game, uh that means, for example, you need to uh like do some uh I would say you need to really run right, and this means the the apps.

C

It will need to collect your like the GPS location like the speed it needs to do some calculation and it's need to finally report such data and uh to to finally give you the rewards on chain. So so this means the this is the data. That's uh decentralized me to process yeah, but let's ask for the question is first, can we can we really do such data processing in a smart contract and actually I think the answer for us is no and also I, I. Think, like the we will introduce later like the reality.

C

Is that uh the like these app developers they will not only need the to use the smart contract. They will also like need to use some centralized uh cloud service like the service from the AWS and the front Microsoft, and the reason they cannot do. That is due to the limitation of the current blockchain or the limitation of the current smart contract. For example.

C

The very typical execution model of the current smart contract is something we call the Unchained execution model so like the the users, they do not see the the real workers that backed the blockchain. Instead, they they, for example, when they try to need to do some calculation.

C

They first submit the transaction and brought courses to all the blockchain and then the workers they will collect the transaction and do the execution for the users and during this process actually, for example like in the Easter room and in the like other chains, a very important limitation of their network is that not all the workers can be trusted right, so maybe like here we have five workers. Three of them are Goods our honest workers. They really do your calculation, but there are two bad workers. For example. This is the bad workers.

C

He deliberately provide the false results, and also there can be some lazy workers, for example, he just not. He just ignored any transaction uh on this chain. He just provides zero for all the execution, so a very important constraint for this network under the current design is that uh the honest worker must be the majority of the network, so actually uh all the workers.

C

They need to do the duplicated execution to verify that the execution result is right and the users they cannot read the result gets the results from any single workers because it can be false, so he have to finally get his results from the Block so because, when the block is really appended to the blockchain, that means most of the workers they have done their verification job. So the result is correct. So this is the current pattern of the smart contract, but this comes the limitation of this.

C

So first is that there is no privacy and I think yeah there. There is already some privacy related projects. Try to solve this problem and also I think yeah. Zero knowledge proof can help solve this problem yeah. But uh this is the problem, at least for the for the January uh blockchain, and the second part is that uh the computation power it cannot be scaled because most of the workers, the all the workers they need to do, the duplicate computation. They need to verify every block.

C

This product is produced yeah, and the third part is that, since the users they have to uh gets the results from blockchain, they need to wait for the latency, the block, a production interval for each block. So at least like six seconds, and if you want small blocks for the for the finalization, then there should be more times to wait and also I think like as a front perspective of a general purpose computation, for example.

C

If you try to store the very large data on chain, it can be very especially like the chain, like is a room. It can be very expensive to do so and, for example, like the smart contract, you cannot keep a long live like network connection from a smart contract because your smart contract can be like executed by different workers, yeah and also it's impossible for the like these workers to delegate the network access for your for your smart contract.

C

So this is all the limitations of the current smart contract, and this can explain the reason why the our developers, uh apart from the smart contract, part they they still need to rely on the centralized service to do the things that you cannot do in the smart contract.

C

So I think this is a very typical, like architecture of of a decentralized, app and I think that's. We have already do a good done, a good job, because you can see we have decentralized a lot of most of the parts of this. uh The apps, for example.

C

We have already have decentralized storage here and all I think all the assets on assets is decentralized yeah, but I think uh the last part to be decentralized is the the back-end program part so now uh people they still need to write a back-end program and deploy them to the centralized web cloud. But this is the problem we try to solve and the uh if I think, the very important specification for the web 3 based backend is first, it needs to be blockchain native uh I.

C

Think the most attractive part from uh like a smart contract is the enforce that execution of the smart contract. So this implies no trust, so no one can like shut down your smart contract or change your logic. So this is I think this is the basis for all the like the trustless infrastructure.

C

So also we try to support. Like uh other uh things, you need to achieve in a back-end programs like the low latency response, like the The Continuous service, for example, you want to do a very heavy, very complex computation in your contract, and it can takes like 10 minutes to go, and we do not want these like computation to be breaking to different pieces, and each pieces is just like six seconds at most yeah, so also we try to as a back-end service, we can see that a backhand, sorry, a back-end service.

C

You need to connect to all the other components in this system, like you need to connect with the storage service, to read and store some data. It need to like connect to the like other chains to like read the balance of our user assumption account and it needs to directly serve the front end, because users may send the request to these servers.

C

And finally, it also needs to connect to the other web 2 Services, because you need to, for example, you want to read the price of of certain pairs, so this is the last part of the connectivity. We also want our backend programs to have the connectivity to connect to all the components of these decentralized apps, and this is our solution.

C

So, first we we have our follow Network, which is a very uh important infrastructure for this, because it has the computing power and we have many many uh servers all over the world and the good part is we all know of them. Yeah and another part, is the fat contract and fat contract defines how you can use our services, because actually we we don't want to like rent the these Hardware to our users and let them do the setup themselves. So instead we try to wrap it into a serverless service.

C

So actually they do not need to care about the underlying Hardwares anymore. They can just deploy their programs and like to deploy it to any uh numbers of the workers they want yeah, but the first part I will cover. Is that how we build this Computing cloud, yeah, so I think the still the very important feature of it is that first, it needs to be permissionless. Is that we want?

C

We want to allow any people who has the hardware to join our Network to share their computer power to the users, but on the other hand, actually we want this Cloud to be trustless is that we do not trust any one of our workers. So we try to provide the like the hardware-based uh promise that our workers cannot like to to peek at our users data to try to provide false results yeah.

C

So this these are the provisioners and the trustless nature are the things we try to provide for our cloud, and now we already have like over uh 17 thousands of the nodes of the world, and all of them are based on the uh the hardware called the security clay. So since many people are here, I think uh at least you have heard like The, Trusted Hardware, so yeah and I I I'll also introduce it for bridge.

C

uh So before we go further. I think uh it's reasonable for us to First Look at the limitation of the existing smart contract and understand why there is such limitations. So the most important reason is that any single worker, often current Network, cannot be trusted.

C

So that means, uh for example, like the Privacy problem, like the duplicate execution, like the block production waiting. All these things is. Is that, because that execution results single executive result must be verified by all the workers uh in this in this chains, and then they can be uploaded to the blockchain and like for these extra functionality.

C

The reason we cannot trust the worker is that we we do not know whether they really do that, for example, they may claim that they have already delegates the network access for you, but they can just do not do it so so all these things are caused by the entrusting nature of the workers. So here comes the problem. What if we can make these workers honest?

C

And here when we say honest actually, it means that all these workers, they will facefully, run the programs we provide to them and they will not provide false result to you yeah. So so this is the honest meaning in our situation and also. This is why we choose the securing claim or The Trusted execution environments as our solution. uh I think fala is not the only team which used the te as its workers, for example the like the recent security network and the secret Network sorry and Oasis lab.

C

They are both the Pioneer to use this de and to combine with the blockchain, but I think the father use it in a slightly different way from all these projects, because the other projects, they mainly focus on the confidentiality promise from the PE, because tea has a hardware-based encryption chip inside it to encrypt all the memory for you, but actually from our side.

C

Since we are a cloud uh confidentiality, it's also nice to have features of it and also it's very important because it's protects the data privacy for our users, but actually for the nature of the of the Cloud. The more important things is that we need to ensure that all the execution of our users program are correct.

C

So that's why we are relying on the execution Integrity promise from the secure claim, and this means the screenplays. They were facefully, uh execute our uh we released program for it and yeah. So so that's the difference of how fala used the T.

C

And to ensure this execution Integrity, we need to ensure that uh first, we need to ensure that every one of our workers they are really running the trusted Hardware. They are running real they're, really running the securing credit and another thing we need to ensure that is that they haven't modified the program we have distributed so and the program is called the period time. It's short for father, runtime and uh actually to to achieve this process.

C

We rely on a process called the remote attestation and the remote testing is a very important component of the modern uh te Solutions and, what's it do, is that first thing in every CPU like the Intel CPU?

C

It has a secret key inside this inside the CP Intel CPU during its manufacturing, and also this means every CPU which is verified by the manufacturer like in this case the Intel. It has the ability to generate a unique remote attestation report called the I quotes, and it contains all the information we what we need to know about this hardware and our program status.

C

For example. This report is first verified and certificate certified by the Intel, so we can know that it's really a valid Hardware manufacturer fandom and also it contains the like the hardware, information and the firmware information of this Hardware. So we can know that it's our program is really running inside a trusty hardware and also we can know about all the vulnerabilities information about this.

C

For example uh like in the Intel report, it will provide something called the security level and the fourth of it will tell you uh we, under the current firmware version, how many known vulnerabilities are there in your Hardware yeah. So this is the. This is what we can learn from the hardware information and also the final part, is that we can. We have some information on the uh the program it's it's now executing inside the sdx so and so from the code harsh.

C

We can know that the like our program, our period time, is not modified by any any workers and also uh like it can contain the initial memory layout of your program. So we can also know that the initial state of our program is determined yeah. So all these things can ensure that every worker in our Network, they are really running the the securing credit hardware and they are really running the program we have distributed.

C

And with this promise, actually we can uh change the previous execution model of the current blockchain and still this is the very typical uh unchain execution model. But in our case actually we turn it into something called option execution. So first here, actually we still have some validators in our chains and actually the consensus is based on the POS. So but in our case the validator, the only uh they are only responsible for the like the transaction collection and the block production, but they do not really do the execution.

C

uh So another very important part in our network is the auction workers and all of them are based on securing claim. So what these workers do is that first, they keep thinking the transaction from blockchain inside our workers and we have implemented the simple payment verification protocol inside our workers. So it can know that the transaction is received is uh valid and during the think process, all the workers they are do the execution of chain and they keep updates their local States.

C

And finally, a very different things is that the users they do not read the execution results Unchained, because we did not post the updated State down chain. They directly read the States from and the results from the workers. So this is what we call the off chance.

C

Yeah and uh another part I will not introduce in today's uh talk- is that about our quality of service Insurance, because uh with the previous protocol, actually, we can ensure, like the security and the privacy of our network, but we cannot ensure that each workers they provide very good quality services, for example, they may reduce the frequency of their CPU and they may like, uh for example, provide very low at the network connection with very low bandwidth. So actually we have another tokenomic to encourage them to provide high quality services.

C

Yeah yeah I think uh until this part uh I think, if you have any questions, we can first have a short q, a so to ensure that I have really introduced the infrastructure of our Network. Well,.

A

Shelby I'll, just sort of you know, add add to the context that I love the characterization, the framing of the problem. The way you guys have framed the web respect, I love the approach you guys are taking I. Think it's very Innovative. It's it's on chain friendly enough, but it has a lot of extensibility off chain. Could you say anything about uh from the perspective of of? um Would you call it a minor or a worker? Would.

C

They have some.

A

Spare compute and they would just run your program and it would run in the background, or what does that? Experience look like for that group of people.

C

uh For workers, actually, we provide uh like all these period time and also like we distributed the final node for them because uh actually in our chains to to be to run as a workers. Actually, first, you need to run nodes to keep thinking the uh like to maintain the P2P connection to the blockchain and also this node.

C

It will keep like thinking that this transactions into the period time so and we all we packed all these things into a Docker image for them, so our workers they can easily uh do all these things with, even if they are, they are not very good operators, so I think to be a workers in our chain is not a very hard job because yeah we have wrapped all these functionalities for them.

D

I I would also like to voice uh how how super cool this is. I know there are a number of folks here, exploring uh teas as well. um One question I have is uh well I. Had several questions, Wes already got to one I have a second half of that question, which is how do you allow someone to detect whether or not they have uh T Hardware before download? Or do you just run? It say you just download the container and then say Oh. We can't detect any T Hardware.

C

uh This is a I think this is can be answering here so actually uh when, for example, when you have downloaded like the file notes and the period time- and you run it- you didn't become our worker, so you have to uh follow the registration protocol. So when you try to register registrate yourself to the blockchain, so the blockchain will require you to Pro, provide a remote testing report and also it will involve like some random challenge inside it to ensure that you cannot replay the previous one. So this means yeah yeah.

C

So your your report here is some user data. So this means the user will include like the random Challenge and also it will include the generative key inside the period time, so we can like. First, we will verify this report and also we can know your identity, which is identified by these Keys yeah.

D

Yeah! Okay, make sense um and then so that you said 17 000 uh workers, how many of those are or or any of them yours or are they all public um people joining and contributing.

C

uh All these workers are public people with this joints. Okay, but uh for example, there is some individuals, but also I, think there can be some people who hosted their own IDC, so one people they can. He can host like a lot of workers. It's also possible in our Network yeah got.

D

It um and I guess uh final question is: uh do you have any case studies or example, jobs that you're out there doing um uh again, I I just this is the coolest I'd love to see how people are actually using it.

C

Yeah sure sure that's what I will cover in The Following part is about a fat contract because the fat contract by by programming the fat contract, then you can use our service yeah. So whatever it's for coming soon,.

D

Okay, oh wait: terrific stuff.

B

There's another question from past here regarding sgx compatibility: what's your view on some of the vulnerabilities uh yeah, the vulnerability is still definitely a Hot Topic yeah.

C

B

C

B

Mean overall for for sgx at the moment we are only supporting sgx, but it's on the roadmap to expand. We don't want to. You know, rely on only one Hardware provider here, um but we also created um or if Dr seven wrote like a whole article on this sjx vulnerability. So maybe, if you want to comment it on it directly,.

C

Yeah I think I can provide some short answer here. So first, uh yes, we are now uh only support sgx, and the next thing we do is that we we are. The thing we are we are doing is that we are trying we are supporting the sgx 2.0, which has a better performance improvements over the first version of sgx.

C

So another solution already on our list is called the AMD Sev so yeah, because these two uh t e Solutions I think they are the the product uh ready uh Solutions in the world now and also I know that there is some open source like Solutions, but uh actually I haven't seen the real products based on them.

C

So when there is really like uh uh such products, yes, we will try to explore, explore the possibilities to whether we can support that part and also another thing uh not like another kind of solution we are looking at is the multi-party computation. uh Sorry, we cannot use the zero knowledge proof, because the uh the model is a little different, because you cannot delegate the execution to others under the zkp so but uh but when the like, the performance of the multi-party computation is improved.

C

Also or I also know, there is many people working on the hardware on that then we are. We are also very glad to come to use that part, because the NPC they also like obey our uh requirements like on the Integrity, execute integrity and the confidentiality, so it also can be used and about the vulnerability yeah.

C

So uh I think most of the vulnerability now on sgx is towards the like the confidentiality because they try to leak some data from the sgx, but uh now, to my best knowledge, the Integrity of the sgx is still prom is still promised. So there is very lit to my personality. There is no thing that can allow you like to control the uh execution Integrity inside sjx, so at least we have some. uh We have the quranius promised and also our system design yeah.

C

We we have some Key Management mechanism to help us to to mitigate this problem, to reduce the attack service from the attackers yeah. uh If, if sulkit was always so it can help like to post my articles on that because yeah many people ask this, especially for the recent secret Network accident, yeah.

B

Yeah, it's in the chat.

C

Yeah, thank you.

C

Yeah, okay, so I think uh uh if there is no more, let's I think we can yeah we can. We can go on like how people can use our Computing, Services yeah, actually I. Think like previous many other uh projects, for example, they choose to migrate.

C

The evm like the ethereum smart contract inside the sgx, but to our part, I I, don't think it's a very good idea because it will, uh even if it's sort of like the Privacy problem, it still suffer from all the other things like also uh I, think it cannot be a little hard to be very complex and the general purpose uh like to implement such background program inside the evm yeah.

C

So that's why we try to re redo all these things and we propose our own programming model called the fair contract and uh the good part is that it's it's very different from the current smart contract and because it's due to the off-chain, computation and every contract is finally deployed to one or more of our workers so and the users they can interact with their with the 500 directly by connecting to these workers and send something we call the option.

C

Query yeah and the things the fat contract itself is executed inside the workers, so, uh for example, if if you like, if the user, they directly interact with it, this means there will be no guests.

C

You just send the request to the fair contract and you get the results directly, and this also means there is no latency so and since the this contract is, is response to these workers, there is no execution time limit, so it can can be executed inside it forever and also since all our workers are uh uh RTE based and also this contract executed inside inside our period time inside the securing claim. So the Privacy is also preserved, and also we enable you to do some cool things directly inside our contract.

C

For example, we support the native. uh We provide the native Network support inside the back contract yeah, so uh so I think the good part for our contrast that first is is still deployed through the through the blockchain. So you need to upload the code in public way inside uh to our blockchain and finally, the code will be distributed to all the to the workers like one-on-one, more workers and I. Think the initial state of this code of these contrast determined because you need to make the code and the like the instantiate transaction public.

C

But after that, after a contract has been finally deployed to the workers, then you can do all the Privacy things, because the following interaction with the contract? No matter you interact with it with the transaction or, like you interact with with the opt-in query, all these things will be end to end encrypted so yeah. So then the Privacy is uh is pretty, is preserved and also we have some mechanism called.

C

The key management which is already covered in my article is that we actually, we preserve some secret key inside our hardware and this key I used to encrypt the states of these uh contracts and also in it's used to encrypt the traffic to and from this contract.

D

I, just um I I asked this in the chat. I'm, really fascinated how you uh allow for arbitrary Network egress, um because obviously that could be you could just create a you know. A bot, Network and I was curious. How you attack that.

C

uh uh First, actually, the like the uh things that the T itself, it's native. It provides native support for networks. So so also we expose such interface to this uh contract and I. Think a very important question here is that how we can ensure that, like uh these workers, they are not like monitoring the the traffic in and out this contract and honestly there's. First, there is some search, a certificates distributed with the appear in time, so actually the workers they cannot inject their own uh certificates.

C

So also, actually, although we say the HTTP support, but in the real case we incur, we still encourage people to do the like to send the https request. So this means no one can see your uh interaction with the like the website and, finally, how we can ensure that these workers, they are not block the traffic from the contract. Then this will fall back to our tokenomic, because we have a high not be checked for them with to ensure that their network service is really good, so yeah.

C

So by combining this part, you can safely use the network uh support.

D

C

The contract directly.

D

I'm asking a slightly I'm asking a slightly different question: if that's okay,.

C

D

um You have 17 000 workers uh if I wanted to DDOS the White House's website, I could have them all make. You know a thousand queries per second against the website and, and that would take it down.

D

um How do you prevent that or or does network access only between you know your your scheduling, note your scheduling cloud and the worker notes.

C

Oh, this is a very good question. So uh yes, so first actually uh by like to prevent the thoughts attack, I would say that you will need a load balance before all these workers, so these loadbacks need to distribute the traffic to all these workers and uh but uh I think uh this means yeah. uh Actually, the final name itself. We do not provide like the Dos uh defense program, but actually we just provide the workers, but I think this works. How, like you manage this work?

C

How you distribute this a traffic uh evenly to all these workers is another question which is should be solved by the current like front-end solution, yeah.

C

D

Not sure whether sorry, that was great I couldn't find the unmute thanks. So much yeah.

C

Thank you, okay, I, I can't and yeah I can continue yeah. So, let's compare like the Unchained smart contract with our opt-in version. So the first is Unchained smart contract. They are really deployed to the blockchain and, for example, may like any one of the workers uh may be lucky enough to be choose to to execute it for one block, but in our case, actually the contract is deployed to one or more often workers and they are there, so they are buying to them, and the second part is how we store these states.

C

The answer is that we do not store the states because, since our workers they are trusted, and only and also we have all these transactions ordered and stored on chain. So any one of these workers. If it's choosed by our like like the if choose from action Governors, then it will be shared a key to decrypt these encrypted transactions and it will replay all the historical uh transactions to get the final States so, and this means the latest States. They are only available inside our workers memory and actually so this means first, the user.

C

They can still send the transaction Unchained and our worker. They will sync these things back to the execution for them and also they can directly send the off-chain queries to these workers to get the first discoveries. You can do some calculation in it and also actually you can use this character like to read the latest States from the chain.

C

So both these things are okay, and the last part is that uh we do not or not only support the unction data operation which just like, like the update of your balance like this, and also we uh support the off-chain operation, which is very unique to us. So, like you can do a SUV request.

C

We have the random number generator inside the hardware and also we provide some volunteer cash inside our workers memory for you to fail to store some data and also you can use the HTTP request to persistent your data like on the solution like the ficoid yeah.

C

So actually, our contract is do not try to replace or like uh any existing smart contract, because I think, for example, uh there is no meaning to implement erc20 contract inside that contract, because the current smart contract can do a very good job. No matter like the iso room odd, like the polygon, they all do very good job by executing the current smart contract. But, for example, if you try to uh the I think the fat contract, the file Network, we meant to replace like the AWS and like the Amazon Lambda.

C

So this is the things we'll try to do so. Also, our show later is that how we enable people like to connect the smart contract to the fat contract.

C

Yeah here is comes some like some use case. I. Think. A very interesting use case is that how fat contract can interact with other chains with other smart contracts, uh since we have HTTP requests and since we have the Privacy promise inside it, but the things we do is that first, you can generate any account of other chains inside the contract directly.

C

For example, I generate a etherum account inside of that contract and if you can transfer some tokens I guess we to it to this account, and this account just serve as the gas fee account for you, and then we we directly connect to the RPC node of our Target chain. So no matter is ethereum. Actually we have already implemented the spot for the evm compatible chains and for the substrate change. So this means our contract had taken directly interact with this RPC node and it will like construct the transaction inside our contract.

C

It will assign it with your Gatsby account and then extend the transaction to this option signal directly.

C

So this This Is How, We Do the cross chain interactions and uh I think uh actually, we have implemented some wrapper on it because uh we want to enable that, for example, an etherum contract it can send some requests and our contract will monitor these request and do the responsible and push the response response directly to the chains to this contract, and this is how we achieve it and we use the BSC as an example- and here we see a very simple like price feed Oracle here so first, we need to deploy uh some contract called The Anchor contract in our Target chains, for example.

C

It will, it will need it's. It's just used to store like all the requests from these uh smart contracts. For example, one small contract will request the price of BTC usdt pair and another. It will request the price for the BTC eth pair and the anchor contract will store all these requests in the queue.

C

Then our contract can directly read the contrast. State use the HTTP request, so it will. It will know that there is such kind of price feed request here and then it's it's really. The price from the like the like the binance, so it can get the result directly and also you can do the Json without parsing inside our contract directly. So it will know the price, and the final thing is it too, is that it constructs the cross chain transaction directly and send it to the RPC node.

C

So finally, this price count to these targets uh program so I think uh sorry, Target is my contract. So this is our way to do the cross chain operation and there is a simple demo for it yeah. So this is the the BSC mainnet and like here we can like a contract, can require the price for the for the trading pair like BTC and usdt.

C

Yeah and this transaction is, is finally packed as and the finalized, and then we can yeah. You can see how these uh requests look like. So first we give this an index. So this is very important because all our fat contract it can, it can run on all these work, mutable workers concurrently. So we do not want like this. This request to be processed like duplicated process, so we give it a unique index to to to to to Market.

C

So we can know that we we only process the each the request of each index once yeah and the the require uh the request is in just encode it, and then we this is our contract.

C

Since this is a demo, we just run the local test net to to show its case yeah and the things that do is that this first read the uh transaction to read the contract in the BSC and the delegate, the request and send the transaction back.

C

Yeah, we can see that this transaction has been submitted back to the BSC and we can check it in these uh exploiter.

C

Yeah, so actually this this, uh this transaction is directly constructed inside of that contract yeah, and it has the like the request, ID, like the the pair and the price, so yeah. So this this such information can be directly processed by the by the smart contract.

C

Okay, uh actually I think with such connectivity. Actually, we can use it to back to many different, like services and try to implement some application, which has to be has not never to be implemented before so another Library we have provided for our developers is called the pink S3 and actually it's uh library to enable our fat contract to directly interact with any S3 API compatible services.

C

For example, uh like we have the forever land which it can be backed by the filecoin, and it provides the S3 API for this service and like so actually, we can generate a key inside our contract directly, and this key can be used to encrypt the user's data and also we can upload. We can do the encryption for our users. We can do upload for our users, so it by with just one simple background: try we can turn any like uh Public Storage service into a trusted or encrypted storage service.

C

So I think there is more possibilities for for us to explore, because yeah just use our best imagination.

C

And the last uh questions I want to answer is how how people pay for our service. So actually we we adopted the stake to compute model, and this means, for example, uh by staking our token, which is called the far plastic in the far you can get the proportional competing power for our Network.

C

So, for example, if you stake, 10 percent of bar then means it means you can deploy a contract to 10 of the workers in our Network and then it they can do the uh like the execution for you and once you do not need our service anymore. You can unstake token to get your token back. So this means there is no fee for the execution of your contract, but I'll say for them.

C

If you really want, if you want to store some data like to our chain, you will still be charged by the like Gately, just like what it is in the Instagram ecosystem, and now we are already launched our close, the beta test for the fair contract, and the next step is that we will release the background support hour minute.

C

Okay, and uh here is some link I, provide to like it contains like the Oracle and like the workshop, and also some inspection to our uh like Fair contract to our like our chains. So you are very welcome like to follow our Twitter or join our Discord for for further introduction, yeah, and that is from from my side, yeah. Thank you.

A

Fantastic, thank you guys so much for the content. This is. uh This is incredibly interesting. Just to see the different dimensions of how you guys are building this out. The economics model is very interesting as well.

A

um I have a couple follow-up questions as well, but I'll just pause and see if anybody else in the call has other questions give them a little bit of opportunity to chime in if they have any.

B

It was a lot of content right and uh thank you again guys for for giving us the whole hour. I know you said it in the beginning: uh it's usually 10 or 20 minutes right per project, so we had a little benefit here today.

A

Yes- and it worked out well because all the topics you guys are discussing are different things that different groups are trying to con to figure out within our own projects. One one technical question I had is a follow-up is within uh the host. That's doing the computation they're going to run the docker image for uh for the fallow worker and that's going to manage all the te and all this.

A

um Is there any uh any safety precaution in there to say if I run my compute job to prevent it from getting access to that workers, machine and finding out what else is going on in the workers? Are there any kind of hypervisor limitations, or is that just the nature of the te environment, the whatever? Whatever job I run, deploying on a fat contract? It would not get access to that Miner's machine.

C

uh Actually now, uh first, the fair contract is finally compiled to the awesome and all the runtime we have is implements uh inside our period time. So actually the things you can do in your contract is limited by our runtime. So now we expose something like the the just. The thing I have mentioned, like the network, says and like the some random generation, because people want it and also some some cash. Some storage, like this yeah but uh I I, think the factor is still very new. It's a very new product.

C

So still, we are very open to like to hear the requirements for our users. For example, if you have some something for them, some functionality you want for the fair contract. We are still open to like to add more functionality to it.

A

Very good and then Zoe one last question might be: if folks want to get their hands, they want to get involved in in the work you guys are doing or if they want to reach out to you, do you recommend any any errors? We should funnel folks to get in touch with you guys.

B

Yeah and maybe we can share the the deck afterwards to other participants or you could distribute it, and then we have all the links so I'm, just joining our Discord directly and happy to leave our emails. Also, if you want to reach out to a more direct um content and um yeah perfect, should have been joining sharing them already the slides here, um but we are also, as Dr Sharon said, like a launching fed contract on a mainnet um next year.

B

We have our closed beta open now, if you're just interested in, like maybe joining there um as an individual. You still have like um some time this week or next week to to send an application and um yeah, and then we are also, of course, working on a grants program right so to support Builders, actually using fed contract and maybe even directly with um with five coinovers, to have like a joint initiative.

B

But um this is just for you guys to know that we really want to incentivize everyone to actually use, of course, the tools and then um provide the benefit for your projects and for a network of our.

A

Very good, we'll make sure to add the all these follow-up notes in the compute over data working group slack channel, so that anybody that comes in here after the fact can get access to all this good information and they can get connected with you guys directly.

B

uh Cool, we should join that tour, I have to invite open yeah and then we can all connect the offline, cool, lovely.

A

All right well, unless anyone else has any questions we'll, let you guys run. Thank you so much for the content. This was lovely and uh more to come.

B

A

Thank you. Everyone.

C

Thank you. Take.

A