SHANGHAI, Sept. 29, 2020 /PRNewswire/ — PlatON, as the infrastructure of the private computing network and distributed economy, is not a pure blockchain project in a strict sense. In other words, the blockchain function is only one part of PlatON.
The vision of PlatON is to become the next generation of infrastructure operator. And the core of PlatON is to price data flow. Everything is designed around computing and data. This is the most basic part of human production and life in the future. Because of the status as as the foundation, the system needs to remain intact. Also, we are well aware of the complexity from research to technology and practice, which makes us extremely cautious.
With the purpose of building the public infrastructure in the digital era, PlatON started the journey of engineering practice. Next, let’s talk about PlatON’s experience in engineering practice and testing in series. First of all, start with the most important index of blockchain — the performance index.
Standardization
Performance has always been one of the most important factors restricting blockchain. Because it is recognized that there is no triangle, it is impossible to guarantee the stable and reliable performance index at the same time. Therefore, the breakthrough of TPS has become the main measurement and breakthrough direction of public chain competition.
The blockchain industry, which is still in its early stage, is in a situation where a variety of schools of thoughts and there is no certain standard for the whole industry. Therefore, it is urgent to carry out the performance testing under the guidance of standardized, complete, rigorous and available evaluation methods. Based on the team’s many years of engineering experience, PlatON has defined a set of standard performance evaluation model and benchmark test process to evaluate the performance of the blockchain system in a standardized way. Moreover, this standardized performance evaluation model has been used in all previous performance tests of PlatON test network.
What is standardized assessment? It must be multi-dimensional, which can be implemented, and has operability.
First of all, through the outermost test tools or test plug-ins in the node, the complete test architecture model set by PlatON injects the workload into the system to be tested in the inner layer, simulates the actual transaction scenario, and monitors and calculates various indicator data.
Secondly, the multi-dimension test indicators are set. The test indexes of the blockchain system are divided into performance indicators, resource indicators, stability indicators, scalability indicators and reliability indicators. The results are obtained from the comprehensive analysis of the performance of a number of index data.
Thirdly, the real test environment has been set up. Blockchain, especially the public chain, is decentralized. Its operating environment is not controlled, and the distribution and configuration of nodes are uneven. Therefore, node configuration should be defined according to the actual operation situation as closely as possible, so as to fully evaluate the operation status of the blockchain system in the real environment.
Finally, PlatON uses the test load which combines basic data, transaction type and request distribution to evaluate specific points in performance space. It also divides benchmark test into five layers, which is a comprehensive evaluation from five aspects: performance, scalability, availability, security and stability, so as to ensure the objective and effective performance evaluation results.
Each TPS is guaranteed to be authentic and reliable.
Winning EOS is just the first step. If there is a standard performance evaluation model, system testing is common. In engineering practice, the test of the underlying system is always the daily work of PlatON. Together with community members, various functional tests, exception tests, security tests, governance exercises and stress tests in various environments were conducted on the test network. At present, PlatON has completed more than ten stress tests in the test network environment and simulation environment, and obtained the first-hand test data, so that we can have a clearer understanding of ourselves and use it in an iterative system.
Recently, PlatON has been conducting a macro benchmark testing. EOS is a public chain with a significant performance at present. Therefore, we choose to use the same test method to do the comparative test with EOS under the same test conditions. The results show that PlatON has achieved a comprehensively better performance in a pseudo-real environment.
In the native token transfer performance test:
The average TPS of PlatON transfer is 9604 transactions / second, and the maximum is 14755 transactions / second.
The average TPS of EOS transfer is 3049 transactions / second, and the maximum is 5393 transactions / second.
The performance of PlatON is better than that of EOS when the CPU utilization is not increasing much, because the parallel transaction mechanism of PlatON greatly improves the performance.
The memory and network resource consumption of PlatON is far less than that of EOS. From the resource consumption graph of native token transfer, it is shown as follows: Although the overall storage consumption in the test process is almost the same, the storage consumption of PlatON is far less than that of EOS on average.
PlatON-EVM performance in the token contract transfer is slightly higher than EOS. PlatON-WASM performance is relatively lower, therefore there still a room for improvement. The consumption of memory network resources of PlatON is far less than that of EOS, and the storage consumption is almost the same.
In the key value contract transfer performance test
In the key value contract called by PlatON-EVM, the average TPS is 5237 transactions / second, and the maximum value is 6304 transactions / second.
The average TPS of PlatON-WASM calling key value contract is 3297 transactions / second. The maximum value is 3797 transactions / second.
The average TPS of EOS calling key value contract is 2451 transactions / second, and the maximum value is 6268 transactions / second.
The test results are as follows: The performance of PlatON is obviously better than that of EOS, and the CPU consumption is higher, which indicates that the utilization rate of PlatON in multi-core CPU is higher. The optimization of consensus mechanism makes the final confirmation time of PlatON much lower than that of EOS.
In the future, we will optimize the technology, continuously test and iterate the infrastructure. We will also cooperate with many institutions to polish the system, so as to maintain the best state after the main network of Trantor launched.
PlatON’s journey of engineering practice will not stop and will continue to focus on the field of data. To construct the data market based on blockchain and privacy computing would bring a brand-new market of trillion US dollars to the world.