On September 6, 2017, the Bank of Japan (BOJ) and European Central Bank (ECB) released their analysis for Project Stella, a cooperative research undertaking that launched in December 2016. The banks sought to “assess the applicability of DLT solutions in the area of financial market infrastructures.” The BOJ and ECB presented three main findings:
- DLT-based solutions could meet the performance needs of a Real-Time Gross Settlement (RTGS) system;
- DLT performance is affected by network size and distance between nodes; and
- DLT solutions have the potential to strengthen resilience and reliability.
Typically operated by a country’s central bank, a Real-Time Gross Settlement (RTGS) system allows banks to immediately send one another low-volume, large-value transactions. TARGET2 is the RTGS for the Eurosystem and BOJ-NET Funds Transfer System is the RTGS for the BOJ. Despite finding “encouraging evidence,” it appears that neither will be transitioning to DLT anytime soon, according to a joint statement by the BOJ and ECB.
“Given the relative immaturity of the technology, DLT is not a solution for large-scale applications like BOJ-NET and TARGET2 at this stage of development.”
The BOJ and ECB found that “a DLT application could process volumes of payments requests comparable to those routed to RTGS systems in the euro area and Japan.” The average traffic of the centralized payment systems is between 10 and 70 transactions per second. A DLT solution could keep pace with the current financial infrastructure but, as one might expect, increasing the transaction requests per second diminished performance.
Adding nodes also caused some latency for “payment execution time,” though the analysis indicated the “resilience” of a DLT network. The collaborative findings showed that a DLT network might “withstand issues such as (i) validating node failures and (ii) incorrect data formats.”
The DLT analysis was performed using Hyperledger Fabric version 0.6.1. ECB used a “virtualised and restricted in-house test environment” featuring two Red Hat Enterprise Linux 7 (RHEL) machines with 16 virtual cores, 8 GB of RAM, and 50 GB of storage. BOJ used cloud computing services with each validating node running on a separate Ubuntu server, each with 7.5 GB of RAM and 8 GB of storage.
Ultimately, the joint analysis expressed “reasons to be optimistic with respect to the capabilities of DLT within payment systems” – a promising indicator for future development.