Project Ubin Phase 2. Exploring Corda, Hyperledger Fabric and Quorum for RTGS clearing/settlement

There are interesting results and findings from this project to explore using DLT for RTGS settlements. In one of my earlier post, I mentioned that one of the uphill tasks is convincing the banks to participate in Ripple's DLT.

In order for banks to work seamlessly with DLT, banks may have to upgrade or overhaul their existing system. That boils down to what choices to make.

1. Continue to use SWIFT which is already mainstream for x-border payments or 
2. using DLT where there is no concrete proof of concept that it can work or create future unforeseen issues

Project Ubin Phase 2 is using 3 DLT, Corda, Hyperledger Fabric & Quorum for RTGS settlements with focus on funds transfer, queuing and resolving gridlocks. Each of these DLT presents their plus and minus features of the DLT which I will list below shortly after viewing their report from Accenture.

There are more than 1 DLT in the market but seldom I see there are comparisons done for each of them at solving real world problems. The report from this project revealed the findings

In gridlock resolution, Corda uses Cycle solver algorithm developed by themselves and Hyperledger and Quorum uses EAF2 algorithm . The EAF2 algorithm is the earliest FIFO-based algorithm used in Germany.

Corda

Extracted from the report: "Corda is a distributed ledger technology platform designed for use with regulated financial institutions. It is designed for recording, managing and synchronising commercial agreements between known and identified parties at scale without compromising privacy"

Plus points

1. To cater to privacy requirements, Corda shares with banks on need to know basis rather than broad broadcast
2. The funds transfer is able to continue even when there is gridlock resolution running
3. High resiliency as a network can still operates even when some nodes failed
4. New banks integration is easy when  adding a new participating node involves only the installation of the new node itself.

Minus points

1. Its UTXO model requires input states to link 1 or more transactions creating a long chains which makes processing slower as time goes by
2. As single  Notary service (Provides uniqueness and/or validating consensus on received transactions by providing signature to indicate transaction finality.) was used, it introduced a potential single point of failure. If the single notary failed, transactions cannot be completed

Hyperledger Fabric

Extracted from the report: "Hyperledger Fabric is a permission network with the ability to set up private channels between participants where each channel maintains an independent ledger. Channel enables information to be shared between parties on a need-to-know basis."

Plus points

1. As this DLT allows channels for bilateral exchange which are private and having their own ledgers, speed in high when it comes to processing

Minus points

1. Difficult to maintain in future as channels will increase with increase participants
2. New payments are queue if there is gridlock resolution running
3. If orderer (An orderer broadcasts the transactions to all peers in a channel for validation before committing the transaction) fails,it presents a single point of failure to network
4. Requires funding to account for each channel

Quorum

Extracted from the report: "Quorum supports both public and private transactions within the permissioned network. The public transactions are broadcast to all the nodes within the network and are processed like regular Ethereum transactions. The private transactions are sent directly to the specified recipients by Quorum’s privacy service Constellation as encrypted blobs."

Plus points

1. If any Quorum node is disconnected from network, the rest of the network can still function as normal.
2. Supports private and public transactions
3. It sends hashes to all participants and if a participant requires validation of transaction in future, they can confirm with the rest of the network by comparing to hashes that network holds without having to trust their counterparty

Minus points

1. Zero-knowledge proofs generation is slow and consumes computing power
2. New payment instructions submitted is set as 'Inactive' and will not be processed during gridlock resolution processing
3. DAPP orchestrated the settlement transaction and if it fails, entire netting round fails for finality stage