Hyperledger Foundation Fabric, 27 Jan 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Fabric Private Chaincode

Description

A virtual meetup hosted by Hyperledger Frankfurt on January 27, 2022 with Maria Munaro, Marcus Brandenburger & Bruno Vavala speaking about Fabric Private Chaincode. Fabric Private Chaincode (FPC) is a new security feature for Hyperledger Fabric that leverages Trusted Execution Technology to protect the integrity and confidentiality of Smart Contracts.

More about Fabric Private Chaincode is at: https://wiki.hyperledger.org/display/fabric/Hyperledger+Fabric+Private+Chaincode

More about this meetup is at: https://www.meetup.com/Hyperledger-Frankfurt/events/282740419/

----
Hyperledger – Open Source Blockchain Technologies

Hyperledger is an open source community focused on developing a suite of stable frameworks, tools and libraries for enterprise-grade blockchain deployments.
It serves as a neutral home for various distributed ledger frameworks including Hyperledger Besu, Fabric, Sawtooth, Indy, as well as tools like Hyperledger Avalon, Cactus and libraries like Hyperledger Aries, Ursa. Learn more about Hyperledger projects: https://www.hyperledger.org/use

Case Studies: https://www.hyperledger.org/learn/case-studies
Training & Certification: https://www.hyperledger.org/learn/training
Tutorials: https://www.hyperledger.org/use/tutorials
Webinars: https://www.hyperledger.org/learn/webinars
Events: https://www.hyperledger.org/events
Vendor Directory: https://www.hyperledger.org/use/vendor-directory

Subscribe to the Hyperledger Newsletter: https://www.hyperledger.org/newsletter

Follow-us on Twitter @Hyperledger

Learn about Hyperledger Membership: https://www.hyperledger.org/about/join

#Hyperledger #Blockchain

A

Okay, you are now live.

B

And welcome everybody to a hyperledger meetup I'd like to introduce the speakers today, who will be speaking about fabric uh private chain code. So with that take it away. Please.

C

All right, thank you, david for for hosting us actually all right. Aha, so you I think you see also the that screen. Let me try to make it a little bit smaller.

C

So can someone say that you still see my screen.

B

C

Awesome: okay, all right! Let's talk about favorite private chance, so I'm super happy that we have the chance today to share this exciting project with you. um So this is um the project is called fabric, private chance. We are a hyper letter uh project. um I mean for uh providing a new feature for fabric um with the subtitle confidentiality um confidential, shank code execution for hyperledger fabric using intel sgx.

C

So today we are three speakers, so it's me marcus, I'm with ibm research and zurich. Then we have maria with home landing pal and then we have also bruno from interlabs, and I suggest that we should start with the motivating so why? Why does this project exist?.

C

Okay and I'll use that somewhere else- maybe here maybe even smaller, great okay, so I I hope that the audience is familiar with blockchain technology and that even we have people here on the call who are familiar with hyperledger fabric.

C

um I think that's also the chance to check if this can be an interactive session, so I would suggest if um someone is familiar with hyperledger fabric, please hit the plus on the chat channel and see if we have some people here actively listening awesome, I see pluses, that's amazing, all right, okay, so ah even more plus is awesome. So I guess I mean, since you know, hyperledger fabric- um I don't know, maybe you have already thought about privacy, sensitive use cases on hyperledger fabric.

C

So what do I mean with sensitive use cases so there I mean we know that um blockchain technology is nice for very different applications and particularly for the verifiability aspect which is provided by blockchain. However, there are um use cases which come with certain security requirements, in particular privacy requirements. Let's take a few examples here. What what I mean with that think about a sealed auction there.

C

We understand that in the field option or a secret option, the the bit secrecy should be maintained during the auction and in the end, an auctioneer reveals the winner of an auction very similar e-voting right. So we understand that in the e-voting scheme or in general voting scheme, we submit our ballot and this ballot should maintain the secrecy right and nobody should ever learn what what was my particular vote, but also health care use cases where, for instance, genomic data analytics um is happening there.

C

We also have a quite good understanding that those genomic data is very uh sensitive data because it it allows to identify uh humans right it, identity. It helps to identify people and in a supply chain use case where multiple organizations they would like to um exchange assets uh for um for some for some value for some money right, but uh certain contracts um should not be revealed in particular, uh people are not really interested in, let's say revealing the price they paid for a certain asset to a competitor, for instance right.

C

C

Luckily, fabric already comes with a bunch of privacy mechanisms built in so, for instance, we can have channels. uh Some of you may already know the concept of a channel, so we can actually split the blockchain in multiple blockchains and only particip. uh Let people participate in the channel uh who need to particip uh paid for instant in the supply chain. We could say: okay, only, the two organizations which are um interact with the contract there on that particular channel that excludes automatically competitors from their channel and from the details of that contract.

C

A second mechanism is private data collections, for instance. I think there was talk about that here on this meetup, I think in in november, so there we learned a little bit more.

C

What that means, however, the to recap: the basic idea is that we use um the letter, the blockchain only to store hashes of the data uh we are exchanging or using in our smart contract, and all the real data um goes between the organizations uh through the data through the through the data collections in some way of of offline communication, direct communication, and then the third thing is transient fields which allows us to add data to transaction without adding it to the transaction.

C

That means we can submit data to the smart contract to the chain code. Without this data ending up in the transaction itself, which is then later committed to the blockchain or stored on the blockchain.

C

However, all these mechanisms- they have a problem- and this is the data visibility at the endorsing peers and the respective organizations. So what do I mean with that? So think about that in hyperledger fabric there are endorsing peers, executing or running a smart contract computing, a result uh then sending it back to the clients, and then the transaction is generated and submitted to the network.

C

However, these endorsing piece there executes a smart contract and in order to execute they need all the data.

C

If we now take our voting as an example back here, um even if the the ballots are in some way encrypted, they need to be decrypted at some point by the smart contract processing them and then there's some computation who actually won the election. That means that the endorsing peer from a security perspective has access to the clear data.

C

If we now imagine that the government would run such an endorsing peer, then we could say well. Maybe this is not a good idea right depending on, if you trust your government or not, but I mean this is just an example.

C

Okay and so, and in order to overcome now this problem of data visibility, we introduce what we call fabric private channel, which is a framework to build, deploy and run private chain codes on hyperledger fabric.

C

The fabric private chain code accommodates blockchain, use cases with strong privacy requirements, for instance. Those examples I just gave, and it does it by executing the chain code or the smart contract by using hardware based trusted execution technology like intel sgx, and by using that trusted execution technology. We basically get three security properties for our application, which uses fabric private chaincode.

C

The first one is confidential compute, which means that all the transaction and the ledger contains only encrypted data.

C

The second is verifiable operations where transactions can be or are verified through a mechanism called attestation, which is something provided by the trusted execution environment. I will come to that also in a second and the third property. Is data misuse prevention, because with this technology, it allows us to bind the data we are using in our transaction to the intended chain code?

C

Okay, let me quickly talk about uh intel sjx, so the idea of enter sgx is that you take your little piece of code. You would like to protect in a certain way and put it in a in the container or in an intel. Sjx speak often called enclave which helps you to reduce the trusted. Compu compute base the tbc. That means that, if my application runs in such an enclave, then we only need to trust, of course, the trusted execution technology, but only the the actual application running inside such an enclave.

C

When we think about the traditional way, how how the trust model in fabric or any other application works, is that we have our application and fabric. This means our peer or fabric, pier, which executes a chain code, and this relies on the operating system. It relies on the trust of the operating system, and the operating system relies on the trust on the hypervisor on the and the hypervisor on the hardware and so on.

C

So if we now, I mean do a computation in that model, we just need to trust the entire box, the entire machine that the execution of the chaincode is correct and that nobody steals my data with introducing into sgx. This picture changes a little bit what you can see here on the right side and it changes in a way that we take now our chain code or a smart contract, put it in an also called enclave which is protected by the hardware itself.

C

This changes our security model in a way that we do not need to trust the peer anymore, but we only rely on the the hardware or the trust in the hardware and the application itself, and by using this technology now we get essentially, let's say three basic security properties. The first one is confidentiality.

C

That means that only the enclave uh sees the data um which it processes. So that means that the peer, for instance, cannot look in inside the memory of the enclave, for instance, and extract certain secrets.

C

The second is integrity, where the hardware ensures that the intended application in our case is changed, is executed as intended, and the third property is verifiability which links to the property. I've mentioned before the the hardware based attestation, which allows a third party to ask the application running inside the enclave. Hey application. Are you running inside the trusted execution technology, and what are you exactly and that process of attenuation will then produce a proof that it executes a cert, a certain application or chain code inside a proper hardware-based trusted execution environment like an enclave?

C

So with this understanding now or this brief understanding, what intel sgx is, I think we can uh look in how we're using this. Actually, I mean, I think you get the the rough idea now we're running our chain code or smart contract inside sjx, for more details on sjx. I would definitely recommend to look on the documentation on the intel website. If you want to learn more about that.

B

Marcus there is a related question in the chat. uh Indirect is asking intel, stops its sgx with its processors right. Would it affect this function?.

C

So, huh okay, so I can answer that or do you want to answer this.

B

I guess I can answer the first part. Maybe you can follow up so and there's it. I guess you're referring to the unavailability of sjx on some client processors and uh well that's the case, but sgx is still available on uh um on server processors, particularly on uh on ice lake, for example, uh where you can have a larger enclave size. So you get more performance, so in sdx is available, marcus.

C

Yes, uh I mean I can try to answer the second part of the question. Would this affect this function? And uh yes, so if the processor does not uh is sdx capable, then we cannot use it in that case right, that's definitely a risk. um However, I think it's it's very interesting because there are, I mean other uh tee trusted execution environment vendors which provide similar technology um right now we fabric private chain code focuses on intel sgx because it fits clearly in our security model, because we are interested in protecting the jncode application.

C

Other te technologies like amd suv, they tar or they aim to protect entire virtual machines, which is which are much bigger than just a chain cut.

C

This is just uh maybe to give you an intuition why we're using intel sjx here, but our in in general, I would say our architecture, which we have not talked about yet because is also applicable to other te technologies other than interstx.

C

But with this let's have a look at the architecture.

C

So this gives you again just a rough, a rough overview or brief overview on what fabric private chain code looks like. So what you can see here on this graphic is, I mean the basic components of a standard fabric network. I mean we have a bunch of peers, we have an ordering service and we have a client or client applications which all together interact and uh allows us to to run applications on top of the blockchain.

C

From the previous slide, we already got the idea. Okay, that now the chain code runs inside an enclave which isolates it uh from from the execute from the the endorsing pier um and through that hardware uh mechanism. uh All the memory con content of that um of the tranquil of that application is actually protected from the peer.

C

So in the this is the first, let's call it the big change. The second big change is that we are um injecting a little bit of additional logic in the deployment phase of of a chance code, where initially, a consortium which runs or wants to use a chance together. They agree on something called the changcode definition and in the chaincode definition.

C

We now include what we call mr enclave, which is a unique identifier of the application or the code running inside the enclave and this vehicle. We are using to enforce that. Only the expected chain code protected by intel sgx is, is used during the lifetime of the smart contact or the chain code.

A

Marcus we have two questions. One is luciano who's asking what would be the advantage of isolating chain code from the pier.

C

So uh I I hope I was already answering that question indirectly with my motivation slide. So the idea is that uh I mean in the blog or the general thing in a blockchain system is that a smart contract is executed by multiple uh participants. So if so, I I'm working for ibm if ibm runs appear.

C

I can trust that organization to to do a proper execution. However, we know that in fabric we can specify an endorsement policy which says: okay, this training code needs to be executed at multiple organizations, but what if I do not trust one of those other organizations entirely, so I would reveal my sensitive data to that organization.

C

Let me bring back that example of an of an orc, often of a voting. If the government runs a voting system, it runs a peer right. I would send my encrypted ballot to that fear of the government and the government needs to decrypt the peer in some way. Okay, let's not talk about. I don't know multi-party computation, homomorphic encryption things like that.

C

That's a total to that, but so the the the key point is that uh you- maybe maybe there are also situations where you trust your organization, but you maybe do not trust the um the host where this peers actually hosted. I mean maybe that appears I don't know uh hosted on a malicious cloud provider.

C

um So there are definitely some cases where you are not fully trusting a peer. I hope that answers that question a little bit and.

A

Then I'm sorry, yes, only one more question to complement.

B

The the other one, uh so this enclave from intel is something centralized where the cinque lives.

C

Well, it's it's not uh centralized in that sense, so every pier would run an enclave and every enclave I mean an enclave belongs to the pier in the end of the of the day, but I mean, as I explained, we have a different trust relationship here, but I mean um the the thing is that we should uh still have the opportunity to send our or our transaction to multiple peers, multiple enclaves to do that. um Yeah. I don't know if you're already familiar with fabric private jet code and then saw something different.

C

So there is so we have some implementation limitations at the moment that our current prototype only supports a single enclave per chain code. But I mean this is an implementation issue, not a conceptional issue.

A

C

Any other related questions to that. I think it's very nice when people are asking the questions instead of typing them.

A

Adam marcus adam made a question says: how do you make sure that the code is run in sgx hardware and not in some kind of emulation layer.

C

Yes, so for that there exists this mechanism um called attestation. What are what I was also mentioning before, which uh um which let the hardware produce a cryptographic proof which you can verify later and that actually plays together with the with the identifier of the actual chain code, as defined in the change definition using that hash of the application, the mr enclave and with this cryptographic proof you received from the chain code, you can verify that and ensure that this was not generated by an emulator layer.

C

For instance, there are, of course, sjx simulation modes, which I mean which cannot provide such a proof, but then there access a fake proof, which you can clearly see that yeah. This is a fake proof this. This does not run in a protected environment. It's not secure, but maybe it's still usable useful for developing purpose.

C

All right um any other questions.

A

No not for now, at least on the chat.

C

All right, so let me now finish the slide and then I will give to bruno so um two more things. So, as I already said so now, we have protected or isolated the chain code from the pier um and in order now to communicate. We also need to establish, let's say, call it a secure channel between the applic, the actual application and the chain code through the pier.

C

That means that fabric private jenkins gives you a mechanism in form of the fpc client sdk, which you can use in your application, which is responsible for exactly doing that. So it's responsible for encrypting and authenticating all the transaction inputs in a way that only the chain code inside the enclave can decrypt and then use the data, and the same actually happens for all the data which is stored on the ledger or in particular on the word state.

C

You know that a chain code can access the word state by performing, get state and put state operations, and there the fpc chain code make sure that this data, which is stored outside the enclave on the word state, is encrypted in a way that only the chain code inside the enclave can later decrypt and access the data and clear again.

C

All right- and I I hope now we have a an understanding so what we can do with fabric private checkout, and with this I would like to give to bruno um to talk about a particular use case in the in the healthcare area, uh and let's see how this is useful, to use all right. So bruno, do you want to take over the share screen.

B

B

Okay, hopefully you can see my screen.

C

So we still see powerpoint, not the full screen version right at the moment.

B

There we go okay, so let me discuss this um uh use case on healthcare and particularly on clinical trials that we developed with um with our partners at soft thread, a startup that deals with works in healthcare, so brief background on clinical trials.

B

Well, probably you are a little familiar with this, um so these are.

B

This is clinical research that is performed to assess the safety and efficacy of drugs and treatments for for people, and this use case was interesting for us because there, because there are several individuals and parties that participate in this um clinical trials and in the end there are lots of regulations and data to be protected and of course, you need some means to automate all the procedures and making sure that security is preserved.

B

Now, let me quickly go through some peculiar aspects of this clinical trials. So, as I said, there are multiple parties and to make a simplified picture, um you can see that there are at least four. You have an institutional review board that takes into consideration and approves proposals for clinical trials for investigators.

B

You have subjects that provide the data that has to be eventually analyzed. The investigators conduct the trial, and maybe some experimenters will take care of actually analyzing the data in order to get to a result.

B

Now there could also be multiple institutions just to keep things simple right now, let's just deal with a with a single one, which is um already complicated enough enough.

B

Now, since you have to deal with the um patient data, uh then uh there are some uh regulator, regulatory compliance to um uh to deal with, uh like, for example, you always have to make sure that you maintain the safety of the participants or in the u.s.

B

For example, you have these hype rules or regulations which mandate that the data is treated according to some standards like that it always has to be protected. There must be access control. uh You have to preserve in the integrity and uh always encrypted. It must be always encrypted when you, um when you store it somewhere.

B

So there are several um um objectives um that must be met in order to comply with this clinical trial. uh With with the with the regulations and one thing that is not shown in the figure, it's the auditing capability, so there is another party, usually public department, that takes care of auditing these clinical trials to check post facto.

B

uh Whether something has happened and then take some countermeasures, maybe something that was happened like in data treatments. In that case, some fines will be issued. So what are the challenges in these clinical trials? um Well, the the big question here is to out how to automate and secure them. Now let me point out that we're not talking about automating and securing the physical collection of data from patients or blood samples, or anything like that here- are talking about digital data in how to automate and secure its analysis.

B

Now, of course, if you are able to automate these procedures well, um you reduce time you save some and you save some cost now for data management and regulatory compliance, things get more more complicated, simply because you again, you have multiple parties to deal with and you have to understand. Where do you place trust whether you place that in the in on the individual or on the institution, and how can you better reduce the trust boundaries in order to um improve security in these cases, just one? Second,.

B

B

The usual, um so the first question that comes up is: how do we solve this problem and usually the first answer is well: let's use a blockchain, and this is what our partners started to started. To do.

B

Blockchain in this case means that if before forms and sensitive data were collected through paper forms, for example, well now they can be easily collected and distributed by using a blockchain such as hyperledger fabric.

B

Now, although this works and brings some benefits because you are really automating these procedures, well, things have not really changed from a privacy or confidentiality perspective simply because before you had uh individuals, trading paper forms and now you have still individual trading pdf forms, um for example. So you still trust the individuals and uh the approach that is being followed from a security perspective is a trust and verify.

B

It means that okay, I'll, give you my data, and I trust you not to do anything unintended, and I trust that there will be some auditors in the background that will control what you do. That will check what you do and with the digital data is basically the same thing but and still you're.

B

Relying on the auditors- and perhaps things are even a bit worse because because by using a blockchain you're, actually distributing this data to a multitude of parties simply because the blockchain is shared uh by default, although it can be permission in the case of hyperledger fabric.

B

So how do we solve this problem in with fpc? Well, the idea in fpc is that you can reduce the privileges that you delegate to um to the parties now here. I have a complicated pictures of um what happens logically, but hopefully I can help you make sense of that.

B

So initially we had direct interaction, interactions between these parties, like the subject and institutions, the institution to provide consent uh for data processing and also that itself or you have the institution delegating the conduct of the trial to the investigator.

B

So the idea fpc is to logically centralize all the steps uh and the verification of the steps uh instead of enclave. But again, I said logically centralized because we are building fpc on hyperledger fabrics. So it's in practice.

B

It is decentralized, so this enclave would essentially contain all the implementations to gather sensitive data from from the subject, as well as any approvals from related parties like the irb for the approval of the of the clinical trial or the investigator to approve experiments that the experimenters are intended to run, and only that, and only when all the regulations are met and the profit checks also are have been analyzed and controlled by the enclave.

B

Only in that case, the enclave will take care of transferring this sensitive data from the experiment, approval chain code to another experiment, always inside an enclave in order to protect confidentiality.

B

So by doing this, you reduce reliance on individual parties and actually increase trust on an automated and verifiable third party agent, which is implemented inside inside the enclave which is supposed to implement all the regulations, regulations and checks that will allow you to eventually increase trust and security on data management.

B

Okay. So this is a picture of the demo that we actually built with with our partners uh again, I understand it's a bit intricate and big, but it is a big and complicated use case. uh Indeed, um uh all these components and and whatever we built you can find that on on our repository on on github. So I will just walk you through these um these components. To give you a better idea of what's going on, uh and then I will show you a video for how it works in practice.

B

So again you can see the different parties um that I showed you before, so the arb that approves the the clinical trial and the approval of the clinical trial and the consent um part from the user. These are have been implemented by our partners and they're, based on plain hyperledger fabric.

B

The big challenges came when we had to deal with with secrets and manage confidential information, and in that case we didn't want, of course, to um to release those to all the peers on the on the fabric network, and for that reason um we wanted to rely on on fpc and so on.

B

Hardware-Based confidential executions or chain codes for for fabric. So what happens here is that the subject will register some data with the experiment, approval chain code and it's a registration with with a promise in this case, and the promise is. That is something like promise me that you will not release the data to parties that have not been previously approved or experiments that have not previously approved so again. That enclave is supposed to implement all the policy and regulations that have that have to be enforced.

B

um The investigator will take care of approving any experiment that the experimenters want to run. So in this case, the subject does not have any a priori knowledge um of what will be the data analytics in in the experiments and that's reasonable, because there may be different ones that can come up and once all the approvals and the registration is um is in process, then the experimental approval chain code inside the enclave can make can determine whether the experimental data so that subject data can be provisioned to the external enclave or not.

B

And if so, it will be provided to another enclave at the experimenter side and the data analysis will happen and it will just release the final result to the experimenter.

B

So let's go through the video here, so we implemented the different parties in this case, particularly the data provider, which is the subject an experimenter which will run the analysis and the principal investigator who was to approve the software before being used.

B

Here we have some patient forms. I will show you the what they contain. That will be sensitive data.

B

We have a diagnose function, which is the data analysis that the experimenter wants to run and I consult on the site that just to keep track of what's going to happen.

B

So this is the the patient form is a standard pdf with the sensitive information like the name of the patient birthday and some questions that the user has to answer like the temperature. Whether now now is occurred, and so on, and the point of this pdf form is that eventually the experimenter has to understand whether whether this subject has a high probability of having this nephritis of renal pelvis origin. So this data will have to be analyzed.

B

Okay, so let's play the part of the data provider or the subject so the subject: uploads the data to our blockchain, select some options during the upload phase.

B

What happens is that the data is encrypted first and the secret encryption key is registered on the fpc chain code, while the encrypted data is instead uploaded to an external database. This is really an optimization simply because the blockchain is a an expensive database, and so you don't really want to use that to upload arbitrarily large data.

B

Okay, once this is performed, let's play the part of the experimenter experimenter decides that he wants to run an ml machine learning, inference model over the sensitive data, and to do so, he decides to use this diagnose python function which will use the patient data to make the inference now. The important part here is these data items. These data items are the variables in the code that will contain the sensitive patient data, but, of course, the patient data still has not been disclosed to to the experimenter or to this function. For the analysis.

B

Okay, so the experimenter submits a description of this experiment, together with the code to the investigators for um for approval, the experiment. The investigators are notified, that there is an experiment pending approval here. Three of them are alerted uh of the new experiment, but according to the policy that we implemented, a single approval is sufficient to for the experiment to be approved, so the investigator has access to all the information about the experiment, the description and particularly the code that is able to to review before making a determination.

B

In this case, everything is fine, so the investigator goes ahead, approves the experiment.

B

Now the fpc chain code considered considers that as completely approved and the experimenter is ready to launch an instance of this experiment.

B

So what are you seeing here um is an instance of the other enclave, the second enclave, uh the one that contains the experimentation service, and this enclave is the one that contains the experiment code and it's the only one that is able to to interact and authenticate itself with the fpc chain code for the sake of making sure that the experiment was approved and so that this enclave is eligible to receive the sensitive patient data.

B

So there is a public key for authentication and also for establishing a secure channel. The experimenter runs. The experiment receives the patient data, the enclave grabs that from the external database and eventually it is able to get the final result, which is 19, and this is the probability that that particular subject has um these of having these the fridays of reno pelvis origin and again, this is the only information that is eventually released to the experimenter.

B

This is because it's how we implemented that, and so the experimenter did not have access to any patient data, so we were able to preserve the confidentiality from uh from when the data was initially provided to the experiment. Approval chain code to the fpc chain code up to when it was eventually analyzed without ever releasing that okay. So this concludes the experiment.

B

A

Case we have um two questions. One is from neo who says in this use case. Does the subject need to hold a private key for transaction signature? If, yes, may I know where is the pk held and does the intel sgx have access to subject private key? May I know if it performs inside sgx.

B

uh Okay, so uh yes to both um yes to both and uh the the reason is simple. The subject has to, of course, authenticate itself in these in these procedures. In the end, it's a client contacting an fpc, an fpc chain code, and since this is based on fabric well, we have a fabric, client and a fabric chain code in the background. So for the sake of authentication, yes and those keys are held locally at the subject side uh and they are yes, they're used to um essentially um trigger transactions on on fpc.

B

uh I would also add that these public keys should also be registered with the chain codes simply because the the fpc chain code has to authenticate or verify that the subject, uh indeed, is a subject enrolled with the institution for that clinical trial. Now, the second part of the question is: uh does the enclave.

B

Use some public and private keepers, and the answer is yes and of course yes, because it has to perform some digital signature of what it's computing and also it has to authenticate itself with with other parties, because parties suggest the subject has to make sure that it is uh talking to a legitimate enclave. So these enclaves, these keys, are generated by the enclave at startup time and they are attested through the hardware-based attestation mechanism that marcus quickly introduced introduced before so.

B

In the end, the chain of trust that you uh will eventually uh verify you know in order to establish trust in those keys is that you will verify the hardware. The hardware will tell you that it is running a certain piece of software, which is this experiment, approval chain code, and this piece of software will include those public keys inside inside this at the station. So essentially you refine the whole chain. You establish trust in those keys and eventually you can use even the private key inside enclave to sign any statement.

B

Other questions.

A

No, the other question marcus um answered it on the chat.

B

Okay, so again, this demo is available on on our repository talk to us. If you need additional information about this or about how to run fpc or order them itself, and I will stop sharing.

B

And maria can take over.

B

C

So, while we're switching here, the the gears I mean so our repositories on github, it's called fabric private chaincode. um We also show the link later here in our slides, um but I can just answer this question. While we are changing things fabric, private change, just google, it you will find it first reverse it.

A

So, just to do a bit of a recap: okay, this is the architecture and what are the components that we have to develop? So there are two components: the fpc chain code and the application that is accessing and calling that chain code right. So how do you develop both of those?

A

The first one is the chain code. How do I develop the chain code, so the chain code? You have to develop it on c plus plus. This is because we are, since we are running in a first execution environment um that is using sgx.

A

This is a requirement of sgx to develop in c plus plus now we have provided a shim to be able to handle a lot of the logic that is required to do the encryption, decryption processing, and all of that we are actually that shim is on top of the current fabric chain code um functionality. So it's. This is not completely different from what you can get with the uh fabric chain code.

A

The only difference is that here when you, for example, hear you call the get state or the put state, you will not have to encrypt um the uh the parameters before, because uh that will be done when the when the invocation is made okay, so you only have to import our shim as with any chain code. You have to develop the the invoke, and this is where you would program your the chain code logic.

A

Now, if we move on to the client side, we need fabric, provides a fabric, client sdk and go so as with the chain code we develop on top of that sdk and since the fabric um client sdk is in go, we, the fpc, client sdk, can also be in go um these. This sdk manages not only the the interaction that we have like here, that we have a submit transaction and an evaluate transaction, but also the life cycle, so all the functionality to be able to implement, deploy um a chain code into an environment. Okay.

A

Here we can see that we get a contract. We submit the contract here, we can see the values again. That is because the um the shim is is the one that is doing all the transformation, and here we have an evaluate transaction okay, and how can we say so that's the theory of how everything is done now. I want to get my hands dirty. How do can I do that?

A

Well, this is our the link to our github repository, where um you're going to be able to find all the things that you need, and if you, if you don't, you can ask us or you can post and we'll try to help you out right. The important thing is also that you're going to say, and um one of the first questions was intel doesn't is not um using, is not um continuing with sgx and bruno said. No, it's not doing it with the client processors.

A

So, if I want to develop, I don't have a big spice link server where I can have the sd. How can I do that? We provide dockerized development environments that can simulate that sgx. So if you've got the the hardware well better for you, we still do provide a dockerized development environment. So you can run everything in there and make it more simple for you and if you don't doesn't matter, you're going to be able to develop the same way because you're going to be using sdx simulation.

A

Now in our repository, we have different samples. Okay, the the simple the simplest sample is the hello world tutorial that will help you not only develop the chain code but also develop the fabric um client to be able to access and execute that chain. Code. Okay and one of the questions that was asked on youtube live was, I want to add it to my network. How can I do that? One of our samples is hey. Remember the the fabric sample of the how to create a test network.

A

Well, we show you how to add fabric private chain code to that sample. Okay, so that way, you okay, you learn how to do with the test network. You create your own network, we'll follow the same rules and just add the bit that we have added so that you can do that, and it's not only with the test network, but also with kubernetes and bruno just presented.

A

The um irb demo, and something that um I would like to add to that is he mentioned that it is in our repository, um but also that there we are using a fabric, smart, the fabric, smart client. What does this mean? He showed all of the different participants, so that network is not a simple network and if I'm I'm a developer- and I have to set up all of those peers to run on my um laptop, it's going to take me a long time.

A

Okay, because it's not already set up, I have to do it, so we piggyback on top of fabric, smart client that provides a programmatic logic to be able to define that network and get it created, and that simplifies the development process. It also has the ability we've integrated it so from fabric smart client. You can also call our utilities our chain codes, and that way it makes it much simpler to be able to develop a a solution to whatever your problem is, and I've tried to. We've got two minutes left for um questions.

A

I wanted to take you guys to the github repository, but I think it's going to be much more useful if we open this up for any questions. If, if you're on the zoom call, you can open um your unmute your microphone and ask a question or if there's already a question in the chat we can answer it.

C

So there is a a question on the chat um who were just asking: is it possible to interact with other blockchain say I want to build a smart contract on, say, ethereum to give some token to those giving away their data on fabric blockchain? How difficult would it be to implement? I think this is a good question. I I often think this is a little bit beyond fabric private chain code.

C

um Well, I mean my answer, is um I don't know actually um but yeah I mean you can definitely build a fabric private chain code which can uh let's say, process details of of inputs, data you're also presenting to an ethereum smart contract. I guess, but so we do not provide any functionality or any layer which makes let's say our technology stack compatible with, for instance, ethereum.

C

I hope this answers the question at least a bit.

C

Any other questions just just feel free to unloot yourself and just jump in here.

B

Okay, I guess it's 10 and pst at least.

A

I guess we can wrap up just one last question um marcus. I don't know if you want to.

C

Alexander was.

A

Asking about the chain code, life cycle.

C

Yes, absolutely so I mean with fabric private chain code. We are, let's say exactly following the same uh jenkins lifecycle I mean you have to you have to package it. You have to install it, I mean deploying it. You have to approve it. I mean all the organizations they have to submit an approved by my org transaction and then at some point you have to do. I think it's it's called a commit, commit chance, definition, transaction, the only thing.

C

What changes here is that, as part of the chance definition, we add the unique identifier of our chain code inside the chenko definition and with the unique identifier I'm referring to that hash of the let's say: the hash of the compile chain code, the expected chain code. So we add this to the tranquil definition.

C

So this is. This is one piece how it changes. Then we have. uh We have a component which we call the enclave registry, which is implemented as a chain code. uh We have not talked about that yet yet today, because this is nothing we are exposing to the users of fpc, so our client sdk and our entire firmware tries to hide that component. However, it is crucial for security, because when it is, it is used when a chain code actually is started.

C

There is this process of, um or there is one additional call. We add to the life cycle, which is called register enclave, which uh you, an admin of an organization using our sdk, performs at a certain peer which tells the peer hey. Please start an instance of that fpc chain code that fpc instant then does what bruno explained a little bit. It produces some crypto keys. It produces a remote attestation which proves that it runs a certain chain code and it's protected by the hardware.

C

Then this attestation is sent to our enclave registry component, which looks at the proof and checks that this proof contains exactly the same hash, which was defined previously in the chenko definition, and only if that matches then we actually allow the enclave to be registered and to go live in the system. That's one thing we have not talked about today.

C

One other minor thing is: I mean this goes really into technical. Deploying details for fabric private jacket is that we are relying on the external service functionality of the pier in order to uh to spawn a chain code. As you know, from from go chain code or javascript chain code, there, you normally when you install it on the pier you tell the peer, hey, here's the code, please um compile it package it as a docker container, and then there is a docker container running at the pier.

C

Our entire build framework gives you the the tools in order to create that docker container. For you, then this will be shipped to the pier and by using the external service or the chain code as a service mechanism, the the peers able to communicate with this docker container. Where now our fpc chain code is running. Those are let's say in in in just uh two minutes how the life cycle changes, but we really try to keep uh as as close as possible to the normal fabric.

C

Life cycle, um maria told you about the fabric samples test network. If you go to that tutorial on our repository, you will you will see those individual steps very well explained in the readme. You can follow them and I think then it should uh yeah. Everything should be clear how this is different from a normal chaincode lifecycle.

C

All right, I think, we're over time. Are there other questions.

A

There's one last question: is it possible to use fbc to privately store input data on the ledger to be processed later similar to voting? The votes will probably be calculated at some moment in the future.

C

Absolutely so, this is actually the use case for fabric private check code. The idea is that we send our data in a way that it is only visible to the actual fpc chain code. The fpc chain code can then store that data on the ledger in an encrypted form and at some point, when you want to process this data, you can trigger the chain code and tell hey now it's the time. Please do something with the data. The chain code gets the data decrypts it process it and without revealing it to the endorsing peer.

C

That's actually the use case of fpc right.

A

And last just to finish, I want to invite anyone who, like the idea and wants to help out we meet on tuesdays. You can check the the time on the calendar on the hyperledger calendar of when our meetings are and we are open to anyone who to new contributors.

A

C

A

Thank you. Thank you, david.

C

Thank you for everyone for joining and um yeah. Hopefully we will.

B

uh Talk again soon.

C

All right, thank you, guys have a good day. Everyone.

B

Have a good day.

C