On September 5th, at the blockchain POD conference hosted by Odaily and co-organized by 36Kr Group Strategy, Liu Yunqu, Institute of Data Science, Tsinghua University, published a speech entitled "Data-Driven Cloud-Network-Chain Integration: The Next Generation Data Network Provides Block Blockchain Platform as a Service (BCPaaS)".
Liu Yunqu believes that from the perspective of big data and the network, blockchain technology may have the potential to realize data confirmation; the combination of blockchain and big data may give birth to trillions of new businesses; blockchain does not yet have the support of huge Data capabilities; blockchain poses a huge challenge to network and data infrastructure.
Existing infrastructure is difficult to support the explosive growth of AI/big data/blockchain. The Next Generation Data Network (NDN) project,It is hoped that through in-depth research on cloud computing, software-defined network and blockchain network technology, cloud-network-chain integration will be realizedsecondary title
The following is the transcript of Liu Yunqu, Institute of Data Science, Tsinghua University’s speech:
Hello everyone, the title of my speech is "Next Generation Data Network: Data-Driven Cloud-Network-Chain Integration". The Institute of Data Science of Tsinghua University was established in April 2014. It is a multi-disciplinary cross-integration, and a data sharing platform established with the core technology of data authentication and secure multi-party computing. Its goal is to realize a big data tap water pipeline, capable of some asset services and innovative technologies, and the most important goal is to achieve data-driven.
The existing infrastructure is difficult to support the explosive growth of AI/big data/blockchain.The three core technologies of cloud, network, and chain each have some shortcomings, such as outdated protocols, data islands, high costs, poor flexibility, difficult implementation, lack of scenarios, etc., as well as governance issues such as data rights confirmation and massive big data migration. Economic and efficiency issues, these are what the next generation data network (NDN) tries to solve. The next-generation data network realizes the integration of cloud, network and chain through in-depth research on cloud computing, software-defined network and blockchain network technology. Support the application scenarios of big data, artificial intelligence and blockchain at the infrastructure layer.
Cloud network chain integration, this is a big ambition. Can we stand at an altitude of 10,000 meters and see what is the relationship between these three technologies? Cloud computing provides the huge computing power and elasticity required for big data processing. Together with edge computing and fog computing, it provides data solutions with different delay performances; AI such as machine learning benefits from big data and at the same time provides Analysis provides a powerful tool; network infrastructure tries to shift to data-centric scenarios, such as CND, NDN, ICN, and DCN that everyone is familiar with. What is the essence? It is to turn the network from bit stream to data; we believe that the future value focus, that is, from the perspective of big data, is the privacy, ownership and usufruct of data. The data is mine, so why do you use it? Is there any benefit to using it? Is this benefit I should get? This is what we think is the core issue of big data in the next stage.
From the perspective of big data and network just mentioned, let's look at blockchain technology:
First, blockchain technology may have the potential to realize data validation.Let me quote Eric Emerson Schmidt, chairman of Google, who believes that the greatest value of blockchain technology is to realize the scarcity of data, that is, data cannot be tampered with or copied casually. Why is scarcity important? Because he is the chairman, he cares about economic issues, and there is no scarcity value without scarcity.
Second, the combination of blockchain technology and big data may give birth to trillions of new businesses.The ownership and usufruct of data, as well as the assetization based on data property rights, are the focus of future value.
Third, blockchain technology does not yet have the ability to support huge data.This is the core problem to be solved by our next-generation data network. How to blockchainize large data sets is a big challenge, and the industry has proposed potential solutions.
Fourth, the blockchain poses a huge challenge to the network and data infrastructure.Based on the broadcast communication model, distributed consensus poses a huge challenge to the network. The broadcast communication model of the blockchain puts enormous pressure on the network. A perfect network supporting the broadcast communication model requires as many as 6T times as many legitimate configurations. This cannot be achieved by expanding the existing infrastructure and requires an architectural upgrade.
We believe that the network is one of the bottlenecks for blockchain-scale adoption.The performance of blockchain, including transaction scale and transaction delay, has become the focus of attention. The performance of relational databases has strong consistency, and commercial databases can reach tens of millions of TPS. In terms of the performance of mature transaction systems, the system requirements are not so high. For example, Visa has an average of 2000TPS, which is the performance required by actual applications. An Internet trading platform, Taobao can achieve hundreds of thousands of transactions per second, and for each commodity, it may be less than one transaction per second. The point I want to talk about is that in terms of performance, distributed blockchains should not blindly pursue indicators, as they cannot compare to relational databases. Moreover, latency is not the same as transaction bandwidth. Transaction latency is the time of a single transaction, and transaction bandwidth is the number of transactions per unit time. Low latency means high bandwidth, but high bandwidth does not necessarily mean low latency. For practical applications, latency is important.
So what is the limitation on the delay?One is the display limit of the code, such as block size, confirmation time, and average transaction volume. The confirmation time itself is a major factor of delay, and the current research shows that it can reach the order of 100ms; the second is the implicit limitation of the network, such as the broadcast communication model of the blockchain, which requires N square bandwidth, thus limiting the block size. scale. The size of the block is limited by the verification computing power. The current 1M size of Bitcoin has been carefully considered; the third is the size of the support node, which is subject to the above-mentioned various restrictions. In scenarios that require fast or a large number of transactions, it can support The number of nodes also becomes smaller. Hierarchical deployment of super nodes may help solve this problem.
The next-generation data network can support high-speed and massive blockchain transaction throughput. The first is the wide-area high-speed broadcast bypass: high-speed wide-area pub/sub service, calling the pub/sub service in the application code, the interface is simple; secondly, it supports sub-second transactions, about 600ms. The 5g core network promises an end-to-end delay of <1ms, so it is not impossible; another is to be able to support large-scale blockchain applications. We provide blockchain communication proxy nodes and broadcast agents at the infrastructure layer. Everyone goes to broadcast.
The next-generation data network can provide blockchain platform as a service (BCPaaS), free operation and maintenance, sub-second transactions, flexible smart contracts and DApp deployment, and provide a unified, complete, and carrier-grade blockchain for the national enterprise alliance chain The operation platform deploys nodes in hundreds of professional ISPs and IDCs across the country.
Thank you for listening to my speech!
