Hyperledger Fabric is continually evolving technology, thanks to its active community providing new features and functionalities. However, it still lacks a user-friendly interface that makes its adoption immediate and straightforward. In particular, one of the main barriers for non-expert users is the comprehension and creation of a network. In this paper, we present FabNet, a user-friendly wizard reducing the e ort for the con guration and deployment of the Fabric blockchain network.
The rapidly growing of interest in blockchain technologies made IT specialists
(es. developers and system administrators) approach this new topic. However,
also due to its novelty, they often nd this technology hard to learn and master
[
Fabric Composer1 instead, was born with the idea of providing a high-level environment abstracting from all the low-level details to focus, for example, only on smart contracts implementation. However, the way used in Composer to write chaincodes and to interact with the network through the Software Development Kit (SDK) changed with the new Fabric versions, making the application deprecated since 2019. With our work, we focus only on the network le system, without involving chaincode or APIs, that could make obsolete the entire system with the rst change.
Our objective is to address technological di culties for developers and system administrators. The rsts are end-users that want to create a network for various purposes, without entering in low-level operations, like the creation and the execution of scripts and les. The seconds are developers that instead need to learn how to build a network, in this case, we provide an easy way to accelerate this process, saving a lot of time in the initial approach that, otherwise, can result very complicated.
Already existing payable solutions, such as Blockchain-as-a-Service (BaaS) provided by companies like Oracle2, Microsoft3 or Amazon4 not always provide a Fabric solution, in addition, the network is deployed directly in their cloud. What we want, is to abstract the con guration and the deployment phase giving also the possibility for developers to enter directly inside the system, modifying les or moving the infrastructure on another space, without remaining close to the bought service.
For these reasons, we present FabNet, a free tool useful to con gure all the network components just using a high-level interface. This makes the creation of a customised network easy also for those developers and system administrators that does not have experiences in this topic. An easy network con guration is useful especially for starting a project without losing time in undesired con gurations. Indeed, sometimes they are not even needed when the main objective is to work on other components like smart contracts and applications. Moreover, having under a unique interface the required parameters can help new developers to understand better the theory and the correlations between the con gurations and the generated les. What we provide is a tool capable of deploying both a test and a production network, that automatically creates all the con guration les and the scripts needed to deploy the nal network.
The rest of the paper is organised as follows. Section 2 gives an overview
of blockchain technology and in particular of Hyperledger Fabric. Section 3
describes the project structure while section 4 shows a short demo of FabNet tool.
Finally, Section 5 concludes recapping the motivation.
1 https://hyperledger.github.io/composer/latest/
2 https://www.oracle.com/it/application-development/cloud-services/
blockchain-platform/
3 https://azure.microsoft.com/it-it/services/blockchain-service/
4 https://aws.amazon.com/it/blockchain/
Blockchain technology was born in the rst years with the concept of
cooperation between dislocated entities, ensuring a high level of trust between the
parties, and provides an immutable way of storing data. With the rst
permissionless blockchains (Bitcoin, Ethereum) it is possible to make exchanges of
assets between its network members. These interactions take place by means of
transactions, stored in the form of blocks in a chain along with a timestamp.
Each block contains a limited number of transactions and, among other
information, the hash of the current block and the hash of the previous block. Data
are maintained within a distributed network of mutually untrusted peers, that
will be in sync with data changes request, thanks to the consensus mechanism.
In the last years, blockchains have adapted to other contexts, with the support
of features provided by the so-called smart contracts. These can be considered
special programs equipped on each node of the blockchain, which de ne a
transaction protocol that applies the terms of an agreement, between one or more
parties involved, enabling auditability between parties [
Hyperledger Fabric [
This kind of system allows to provide exibility, scalability and privacy in contexts that require these features. The absence of a mining mechanism enables a fast validation and con rmation of transactions. This is achieved thanks to the consensus management system, that allows con guring an arbitrary consensus algorithm, taking into account the requirements of the system intended to implement. Indeed, the arbitrary consensus and the permissioned nature of Fabric guarantee a faster protocol respecting to the Ethereum Proof of Work.
The privacy aspect concerns the con dentiality of transaction and data. This is achieved in Fabric by using its channel architecture and membership service to restrict the distribution of con dential information exclusively to authorised nodes. This is achieved de ning channels inside the network in which only a set of chosen nodes can participate. Nodes and channels are also regulated through policies. Consequently, channel's ledger is only accessible to its members and the channel's organisations must approve each peer's membership to the channel. Authentication and identity management are guaranteed through a exible infrastructure based on the PKI cryptographic scheme and Certi cate Authority, which facilitates the joining of a new organisation in the private network. Also, Transport Layer Security cryptographic protocol is used to provide communications security over the network nodes.
The Fabric network is made by di erent components5, reported in Fig. 1 and described below.
Peers are the nodes grouped into organisations, de ned as trust domain in which a peer trusts all peers only within its organisation. Peers execute and/or validate transactions, and maintain the ledger. The group of de ned organisations participating in a channel is called consortium. The ordering service, composed by a set of Ordering Service Nodes (OSNs), establishes consensus and atomically broadcast state updates. The ordering is stateless and decoupled from the peers, it does not take part in the transaction execution and validation process. The Membership Service Provider (MSP), maintains the identities 5 https://hyperledger-fabric.readthedocs.io/en/release-2.2/key_concepts. html of all the nodes (clients, peers, OSNs) inside an organisation. It comprises mechanisms for authenticate transactions, verify the integrity of transactions, sign and validate endorsements, key management and registration of nodes. Smart contracts with system chaincodes de ne the transaction logic and the blockchain settings. They are de ned in the channel and stored in each organisation peer. The ledger maintains the transactions history, there is one ledger for channel which copies are stored in the organisations peers. In addition, a snapshot of the most recent state is stored in a key-value world state. Finally, the administrators have permissions for di erent operations, like creating and assigning peers, creating network con gurations les and modifying policies through system les.
In this section we present the FabNet tool focusing on its architecture and interface. The tool is composed by a SpringBoot back-end in Kotlin and an Angular front-end, available to everyone at http://pros.unicam.it/FabNet/. The main task of the interface is to take inputs from the user and to pass them to the server. These inputs are described in Tab. 3, that contains all the parameters that the user has to ll in order to generate a con guration. From these parameters, the server automatically creates the les needed to create a network. In particular, these les are 51, counting les, scripts and sub-folders; they are related to: 1. Certi cations and keys; 2. Peers and orderers containers; 3. Channel pro le, de ned by di erent sections: a) orderer; b) consortium, de ning the one creating the network; c) consortiums, de ning all the consortiums in the network; d) application; e) capabilities and policies.
These les are lled up with the information passed through di erent parameters in the interface, listed in Tab. 3.
Once the con guration has been de ned, the client will send it as a JSON le to the server, which will generate the related les, returning a zip folder containing them. The sequence ow in Fig. 2 recaps the steps regarding network creation.
It is important to notice that with our approach the user does not have to create manually all the described infrastructure. However, once downloaded, the user is free to modify all the desired parts without any restriction.
Once the system folder is generated, to deploy the network is necessary to launch the autorun le. This le contains and automatically executes all the commands Network name required to deploy the nal network, lifting the user from other operations. In particular, the automatised operations for a total of 14 commands can be synthesised in: 1. Download required binaries 2. Start CAs and creates artefacts for each organisation; 3. Create genesis block and channels txs; 4. Start Docker containers for each org; 5. Link the started containers so that they can communicate; 6. Create the channel; 7. Join peers to the channel.
Notably, the single le is used to deploy a local network, for remote peers, some additional steps are required. Indeed, in a distributed context, each organisation should extract, from the generated le system, the folder regarding its org. After this, it is also necessary to run the CA script generating the artefacts. However, we are currently working on a FabNet extension to provide the GUI also for the creation of a distributed network. In this section we show a demo of FabNet using a basic testing con guration, describing the ow and the tool functionalities linked with some screens. All the steps described in the following subsection can be replicated using the links at http://pros.unicam.it/FabNet.
The tool is composed by 4 main panels, the rst one, reported in Fig. 3, allows to specify the network name and the organisations involved (name, domain); it is also possible to add other organisations and to import/export JSON conguration. We introduce a brief scenario involving two organisations, org1 and org2, that want to collaborate in their network. The con guration used in this example is available at http://pros.unicam.it/test-network.
In the tool there is also a panel related to organisations, where for each of them the user has to insert certi cation authority information and can handle members. In particular, it is possible to add or remove peers, clients, admins and orderers. Depending on the member type, di erent information is required, like host IP and ports. As soon organisations and network are de ned, the consortium must be created. It is possible to create 1 or more consortiums, de ning the participating organisations. Finally, it is possible to specify the channel in the network, requiring a consortium and, optionally, external organisations.
If the network con guration is made with local peers, to deploy the entire network is necessary to run only the start.sh le (download at http://pros.unicam.it/start/). Instead, if the organisation's members are dislocated at remote addresses, each user has to take his related folder and run the inside scripts. The main focus of our future extension aims at removing also these steps, automatically generating one folder and one start script for each organisation. Fig. 4 shows the network participants once deployed, in particular, Fabric network uses Docker containers to run organisation members. In Fig. 5 instead, the status, addresses and names of members are displayed.
The previous network scenario showed a simple scenario with two collaborating organisations but, more in general, FabNet can be used in more contexts. Firstly, it can be used by di erent type of users with di erent purposes, like developers that want to approach this new technology or for users that want only to create already usable artefacts.
Also in inter-organisational environments, like the ones analysed in [
Think about companies that decide to work together, creating also a blockchain infrastructure, in this case FabNet suits also the need of prototyping networks. Indeed, in the rst phases, designers have to collaborate, modelling the organisations communication and thinking about structuring the network, handling peers and all the other components. In that moment, a tool allowing the creation of more networks for testing di erent con gurations could be very useful.
In this context, developers can work together in a sandbox environment, changing con gurations if needed without e ort only by creating and running new networks.
In this work, we presented FabNet, a tool that aims to help IT specialists to automatise the con guration and the deployment of an Hyperledger Fabric network, thanks to the user-friendly interface that allow con guring les without manually entering into each of them. Our contribution mainly targets novices in the blockchain technology, in particular when they have to deal with the creation of a Fabric network. Indeed, FabNet doesn't require any e ort for con gurations, delegating the creation and the lling of les to the interface, and automating the generation of artefacts. We also contribute, providing a tool capable of creating and deploying advanced production-ready networks without involving the user with additional tasks. FabNet supports a large scale of usages, starting from benchmarking applications until development ones.