Scalable communication middleware for permissioned distributed ledgers

Abstract

Distributed Ledger Technology (DLT) is rapidly emerging as a new paradigm for automating complex business processes in secure and decentralised fashion. Currently, however, its wider adoption is hampered by scalability problems [3] rooted in an inherent tension between stringent consistency, security, and robustness requirements on one hand, and growing application demand coupled with high performance expectations on the other. For example, popular peer-to-peer DLTs based on proof-of-work consensus [4] can only improve the transaction throughput by degrading their security and consistency guarantees, which is unacceptable in the enterprise and mission-critical settings. To improve scalability, recently proposed permissioned DLT systems utilise stronger trust models to replace slow proofof-work consensus with more efficient Byzantine fault-tolerant (BFT) replication protocols [2]. Specifically, the reference architecture [1] introduced by the HyperLedger Project (HLP), delegates the task of running BFT to a small trusted core of dedicated entities, called orderers, and ofloads the management of the application state to a separate set of entities, called peers. The orderers assemble client requests into transaction blocks, and put them through the BFT consensus to establish a global total order; the peers process the incoming ordered transaction blocks, and apply their constituent transactions to the application state. All entities are connected via a reliable broadcast layer, which is utilised by the orderers to propagate ordered transaction blocks to the peers.