For instance, the world’s best-known blockchain network, Bitcoin, can only confirm an average of about 3-4 transactions per second, whereas, by comparison, the Visa credit card network can handle something like 20k transactions per second, a major gap in performance between the two systems.
However, centralized systems offer better performance by sacrificing security, interoperability and inclusiveness, all of which are hallmarks of some of the best blockchains out there, such as Bitcoin, Ethereum, and Indigo.
Even among the different breeds of blockchains, performance varies widely. For example, Ripple, the self-proclaimed “enterprise blockchain solution for global payments”, announced on Twitter that it was capable of processing 1500 transactions per second, seriously shrinking the performance gap between centralized and decentralized systems.
Because of all this, the Stratumn team made the analysis performance in blockchain-based systems a key area of focus in our ongoing research and development work.
However, to carry out a fair comparison, we must include performance factors beyond the ones specific to centralized systems (network traffic and data storage efficiency). Thus, we included analysis related to network structure (topologies), reliability, and governance, all of which are key differentiating factors in helping us decide when and where to utilize which kind of technology.
With these goals in mind, we are pleased to announce that we have reached the first major milestone in our research work and are excited to share the results in the paper, “Analyzing Performance in Blockchain-Based Systems". In this work:
We propose a methodology for analyzing performance in blockchain-based systems, evaluate the existing tools in the space, and define a set of software requirements for a new performance benchmarking tool. We invite input and participation in our ongoing efforts from the wider software development community.
Analyzing Performance in Blockchain-Based Systems,
Anuj Das Gupta & Andrew Dickinson
Read the full paper, here.