Scroll to top

Decentralized matching engine testing on Hashgraph


researchhub - June 20, 2018 - 0 comments

One might say that the Hashgraph is the Blockchain killer. Let’s check it out. Hashgraph allows building distributed apps with near to real-time consensus. For instance, imagine you trade cryptocurrencies but without having a centralized matching engine. Rather than you have your own instance of the matching engine on your local machine that is in sync with many other instances in the network.

Lykke Exchange runs this competition to build the best test case scenario that would show how the Hashgraph consensus will perform in the highly loaded environment.

Decentralized matching engine app is required based on the Hashgraph SDK:

  • Multiple orderbooks with each-to-each tradable assets (BTC, ETH, USD, EUR…)
  • Trading with limit orders
  • Continuous double auction
  • Price-time priority matching

Orderbook state and balances should be in sync with multiple nodes. Sending an order triggers orderbook updating. Several thousands of nodes should emulate trading activity. Each node should generate at least several orders (or canceling orders) per second. Testing results visualization is required.

Some metrics to be measured:

  • Latency for the consensus-based orderbook for each node
  • Average execution time for trades
  • How are these metrics correlated with the number of events in the network
  • etc.

Information about Hashgraph could be found here: https://hashgraph.com/

SDK download: http://www.swirlds.com/download/
Simple matching engine demo is inside SDK

Winner

Alexander Daurkin

See the competition’s description and results on Lykke Streams.

Related posts