Blockchain system that can support heavy application load and is very fast when the load is light

“

This work uses a strategy called "Parallel Chains" to improve the throughput of the system, at the cost of increasing the end-to-end latency in transaction processing. With n chains, the throughput achieved it O(n) while the latency is O(log n), so asymptotically this approach is a good tradeoff. Our solution seems to be a clear improvement over Blockchains that use the "Parallel Chains" to improve throughput, however, some like Prism use the chains improve the latency and other mechanisms to handle the throughput. This alternative throughput improvement is to separate the dissemination and the ordering of the transactions, and has more recently been adopted by influential works like "Narwhal and Tusk" . This does increase latency in the processing of transactions like "Parallel Chains" but it's harder to measure its impact. Currently, the research focus in Blockchain scalability is in adapting classical BFT architectures and speeding them to be able to handle the high number of participants that are required in a Blockchain. These solutions do not support permissionless Blockchains "out of the box", like Blockmess can, however using a protocol like "Hybrid Chains" could allow BFT-like architectures to be used in permissionless settings. Some interesting BFT-like solutions are "Mir-BFT" and "State machine replication scalability made simple". These papers improve BFT throughput with a strategy that is similar to "Parallel Chains", with the same drawbacks. Blockmess' ideas could be used to improve these systems.

”