The DLPS: A New Framework for Benchmarking Blockchains

The Energy Footprint of Blockchain Consensus Mechanisms Beyond Proof-of-Work

“…It should, furthermore, consider the network-wide energy consumption beyond validator nodes (i.e., by including all full nodes and auxiliary services) to arrive at a more holistic view of the overall energy consumption of DLT systems. Applying benchmarking frameworks [32] to measure the actual energy consumption might be particularly worthwhile in the context of permissioned systems that aim for high performance. In addition, analyzing the actual hardware configurations, instead of relying on rough estimates, might prove a worthwhile extension.…”

Fairness, Integrity, and Privacy in a Scalable Blockchain-based Federated Learning System

“…Note that the π c operations are significantly cheaper than the π w operations, mainly since the latter requires the vector L as public input and updates w g . As long as operations on public blockchains are that costly, using a permissioned blockchain like Quorum can allow not only to reduce costs but also allowing for considerably higher throughput [74].…”

A Serverless Distributed Ledger for Enterprises

“…Existing permissioned blockchains like Fabric or Quorum are known to be both CPU bound and to expose limited multi-CPU/multi-core parallelism, restricting their ability to scale elastically on demand [27]. Popular permissioned blockchains like Fabric and Quorum also employ databases, such as LevelDB or CouchDB, that have storage limits, unlike our approach's reliance on (the effectively limitless) cloud-hosted NoSQL database storage engines.…”

“…Extant literature has already demonstrated that permissioned blockchains like Fabric and Quorum generally exhibit low latency and fast finality [27] on the order of several hundreds of milliseconds to a few seconds [27] and hence significantly improve on their public blockchain counterparts. While this is already suitable to address many enterprises' requirements, a serverless approach can further improve on these outcomes through the use of massively parallel computation and -at least for intra-datacenter applications -access to CSP dark fiber.…”

“…To examine the performance effect of multi-machine parallelism and access to massive data transfer parallelism available in the public cloud, we compared our approach to two permissioned blockchains, Fabric and Quorum, as these generally exhibited the best performance among several permissioned blockchains in a performance analysis [27]. For our benchmarks, we leveraged the distributed ledger performance scan (DLPS), a standardized tool for determining maximum throughput, latency, and resource metrics [27]. We set up a Fabric network with 8 peers and 4 orderers, and a Quorum network of 8 nodes to compare with an 8-"node" serverless blockchain.…”