Blockchain software for financial applications – Jeremy Serror

Introduction

A blockchain can be understood as a database equipped with specific properties, implemented using cryptography and distributed algorithms. It is typically used to deploy critical applications requiring a high level of availability. Well-reviewed and well-tested open-source technologies are a natural fit: we consider Ethereum, Tezos, Fabric and Corda as best suited for further evaluation by financial institutions. We use the term replicated ledger for Ethereum and Tezos, and the term distributed ledger for Fabric and Corda.

Without limiting the scope of possible applications, we focus on the deployment of a ledger for the exchange of currency, assets or financial instruments: the goal is to establish a value transfer network. The underlying technology keeps official and immutable records, allowing the exchange of units of account in a trust-less environment. Blockchain simplifies and improves the security of applications built overtop: such as currency, payments, asset tokenization, treasury optimization, or OTC trading. Cryptocurrencies technology is directly applicable to such use cases. For other applications, such as the exchange of financial information, the reconciliation of financial operations, or the modelisation of secure business logics, a full replicated ledger (as used with cryptocurrencies) is often not required, favoring the use of a more distributed (or partionned) ledger: some blockchain fundamentals are preserved without necessarily having to reach a global consensus on every transaction.

Ethereum and Tezos are similar technologies from a functionality standpoint, but they differ greatly in terms of conception and technical ambition.

Fabric and Corda allow the definition of more general distributed networks and do not impose a global consensus. While it is possible to configure these technologies to operate in a manner close to Ethereum or Tezos, it is not their primary use case. Indeed, relaxing the need for a global consensus delivers a higher throughput.

RESILIENCY AND PERFORMANCE

Ethereum and Tezos aim to establish homogeneous networks, with a stable and predictable level of performance. Fabric and Corda are comparable to traditional networks of services offering a higher level of performance that can be more complex to guarantee over time.

DEVELOPMENT ENVIRONMENT

Fabric and Corda are akin to time-stamping platforms with authenticated computations certified by network nodes, while their execution is itself not part of the consensus. Tezos and Ethereum place smart contracts code on the blockchain and their execution is fully guaranteed by the network.

SECURITY AND TECHNOLOGY RISK

Public blockchain implementations offer a unique level of software security given their constant exposure to real world threats. Distributed ledger networks, by construction, offer a wider surface of attack and remain difficult to test and audit. From a reliability point of view, the design of Tezos gives it a convincing advantage.

CONFIDENTIALITY

Keeping operations confidential can be achieved in two ways. Off-chain, with external mechanisms built on top of the blockchain to exchange information; or, on-chain, by maintaining information on the ledger. While the on-chain approach remains limited to certain applications as of today, the technology is rapidly progressing and may be a game changer in blockchain adoption.

When independent entities want to create a shared infrastructure of trust without the need for a global consensus, the approach employed by Corda and Fabric is well suited. To maintain a global ledger, and when it is critical to reach consensus on the outstanding supply and on transaction finality, the approach employed by Ethereum and Tezos is better suited. Replicated ledgers are clearly a compromise: deliver the highest stability and robustness at the expense of performance and with additional complexity for confidentiality. Distributed ledgers rely on different assumptions: maintain confidentiality and deliver performance, but with security guarantees that vary greatly depending on the exact application.

Corda and Fabric offer mechanisms that can also be developed on top of a blockchain network. The question is whether this model, which is quite different from the spirit of initial blockchains, is suitable, or if a custom solution is best, in particular if existing services within the organization already offer the required storage and communication infrastructure. The richness and flexibility of Corda and Fabric are also their greatest weakness, preventing the emergence of standard deployment strategies validated by the industry. Distributed ledgers are well suited to handle independent interactions within small groups of entities, and in particular when there is limited coupling and short asset lifetimes. Supply-chain is a good example: tracked items are independent from one another, and after processing of an item, its state remains unmodified.

Ethereum and Tezos are better suited to transactions between a large and dynamic number of entities sharing a common representation of assets with long-term value. Ethereum and Tezos offer software that has survived the constant challenge of securing large-scale public blockchains, reaching a level of safety likely unattainable by enterprise-driven efforts, which only have accumulated experience in the form of disparate private deployments. Smart contracts must be developed to extend the limited core blockchain protocol, however this is done without weakening the fundamental guarantees of the blockchain. Ethereum benefits from an older and richer environment, which is a strength and a weakness. New development efforts are focused on Eth2, raising questions over the long-term viability of the current protocol. Tezos directly offers a dynamically upgradable protocol, which, along with the modularity of its client, should allow for more substantial evolutions than are likely possible with Ethereum. Tezos also benefits from a skilled community of developers, clearly committed to the long-term development of the most robust software possible.