He has a lot of auras on his head: one of the first two founders of IBM Hyperledger Fabric, the IBM blockchain and the first product manager of the super-books Fabric, and the drafter of the Hyperledger white paper. However, in 2016, he chose to come to Ping An Financial Accountant, the unicorn, and opened up a new path in the blockchain field with his own concept of cryptography and distributed system research.
He is Lu Yifan, the general manager of the blockchain R&D department of Ping An Financial Accounts, which is currently valued at US$7.5 billion.
- Viewpoint | Digital currency will open the currency scenario for blockchain technology applications
- He Baohong of the Institute of Information and Communication: "Deviation from the virtual reality" is the overall situation of blockchain development in 2019
- Industry Blockchain Weekly: Banks Take the Lead, Hainan's Policies Are Most Aggressive
- In the five directions of the Central Political Bureau's application of blockchain chain, which companies are doing?
- Oil giant Shell invests in US venture capital LO3 Energy to explore blockchain power platform
- Babbitt launches "strict selection platform for industrial blockchain", 3 major services help technology landing
(In 2018, Ping An Financial Accountant won the “IDC Financial Technology Blockchain Practice Award” issued by IDC, an internationally renowned data company. This is the first time in the blockchain field that the award was won, in the middle of Lu Yifan)
In 2019, blockchain commercialization accelerated, and large and small companies actively prepared for war. As a prophet, Ping An established the blockchain research team as early as March 2016. After three years of painstaking efforts, I am now showing off.
According to the 2018 Annual Report released by Ping An Bank , Ping An's self-developed credit chain (FiMAX) blockchain technology can achieve 50,000 high throughput per second while maintaining a low latency of less than 0.01 seconds. Blockchain nodes More than 44,600.
How is the high TPS of 壹通通 realized in the context of the general exaggeration of TPS in the industry? Far beyond the zkSNARKs 10,000 times the 3D zero knowledge proves what is sacred? Jingdong, Baidu, etc. have opened up the underlying technology of the blockchain. Is the account book ready for open source? BaaS landing is difficult, how to deal with the account? As a big company, do users believe in blockchain technology or believe in the brand of Ping An? Is it really reliable to hash the user privacy?
With these questions, Babbitt has recently entered the financial accountant headquarters and has conducted in-depth exchanges with Lu Yifan.
Babbitt: Ping An is a big company. Is blockchain technology in the subsidiary of Financial Accounts?
Lu Yifan : Ping An has more than 30 subsidiaries. The research and development of the bottom layer technology of the blockchain is mainly on the account book, and also takes into account the development of the business, helping other subsidiaries to make plans and needs. When working with other subsidiaries, some of us provide a complete solution, and some provide a development interface to output in the so-called BAAS mode. The blockchain projects that you see in the market are important to the safety of the blockchain project. It can be understood that we are the core of the core of the blockchain technology in Ping An . We do not object to other subsidiaries using blockchain technology from other teams, such as fabric. After all, we can make what others think is magic .
Babbitt: In terms of blockchain, what is the size and configuration of the account?
Lu Yifan : The team I bring is responsible for the underlying research and development and the overall external output service, with about 80 people . There are many other business teams in the financial account, and each business team has a corresponding blockchain staffing.
(The staff of the financial credit pass block chain R&D department with the title of "the most powerful blockchain team in the universe", Lu Yifan in the middle)
Babbitt: Ping An began to pay attention to the blockchain as early as 2016, but from the outside world, you are still very mysterious?
Lu Yifan : Because of the customer's reasons, some international big projects have not been publicized. For example, the docking of the Hong Kong blockchain trade network (eTradeConnect) and the European trade blockchain platform (We.Trade) has not been reported in China, but there are many reports in the international news media. Some of them are very powerful in the business side, they will take the lead in the development of the conference, such as the Tianjin Port Blockchain Verification Pilot Project in April this year.
We have too many things, no time and energy to participate in various meetings. To be honest, it is a waste of time to attend these meetings . I feel that good projects don't need to be deliberately packaged and they will be packaged.
Babbitt: Ping An’s 2018 annual report shows that Ping An has built the world’s largest commercial blockchain platform through financial account, providing services to more than 200 banks, 200,000 companies and 500 governments and other commercial organizations at home and abroad. More than 44,000 blockchain nodes. Excuse me, are these 44,000 nodes referring to servers?
Lu Yifan : It is an independent machine. One organization usually has at least 4 nodes, which basically meets the local double-active and off-site backup. The particularity of our framework is fully encrypted. All data is encrypted by the participant's own secret key. The entire blockchain network is all ciphertext data, without any form of plaintext. Because of the fully encrypted framework, the control of all data is in the hands of the participants themselves, and only independent systems that can be independent of their own data can be called nodes.
Babbitt: Ping An's self-developed credit chain (FiMAX) blockchain technology can achieve 50,000 high throughput per second while maintaining a low latency of less than 0.01 seconds. Under what circumstances is the above data implemented?
Lu Yifan : In the “Technical Standards for Cross-border Trade Blockchain ” we wrote, in the national secret algorithm environment, the configuration with throughput of 10,000 TPS is: 8-core 2.3GHz CPU, 8G memory. It should be noted that the TPS here is the average value, not the peak value.
The implementation condition of delay of 0.01 seconds: This is implemented in the intranet room, and the network is stable without network delay. The network delay affects the delay time. If the bandwidth is long enough, it will not affect the throughput. If it is in the external network, it will definitely affect the delay, and the impact will be calculated according to the actual physical distance between the nodes.
Everyone pays more attention to TPS, but what is actually pursued in commercial applications is not TPS itself, but how to achieve the highest throughput and reduce the unit cost of TPS with the least machine and the lowest machine configuration . For example, now in the national secret environment to 10,000 TPS requires 8 core CPU, can it be reduced to 4 cores? If the customer has only one CPU machine, can it support 2000TPS? This is the real purpose of greatly reducing the cost of operation and maintenance.
In business conduct, what really competes is your cost .
Babbitt: In the cross-border trade blockchain technology program, which 3D zero knowledge proves that the super zkSNARKs is 10,000 times, how is it achieved?
Lu Yifan : To be honest, the zkSNARKs algorithm has many problems . It is not only inefficient, but its own security assumptions have not been repeatedly and rigorously confirmed. Therefore, many people know about the password circle.
Because we use a full encryption framework, we must be able to do the ciphertext four operations (+, -, *, /) test and ciphertext relationship test (>, <, =). Under the full ciphertext framework, all user data is encrypted by itself, and we must have a mechanism to ensure running. 壹 通 has strong research capabilities (algorithm design) and engineering strength .
The 3D zero-knowledge proof is designed by yourself and is different from any one on the market today. It includes multi-party calculation, zero knowledge, homomorphism and other characteristics. It is not only the refinement of existing algorithms, but also the integration with innovative algorithms. It is a very complicated cryptography protocol, which can achieve the full homomorphism effect of addition, subtraction, multiplication and division in application. .
Our 3D Zero Knowledge Algorithm Paper will be published in a top journal in the near future.
Babbitt: Is the blockchain technology of 壹Accounts open source?
Lu Yifan : For those customers who are particularly important, there is planning in this regard. It is not yet mature enough for public open source.
Babbitt: Many startups do BaaS, but the situation is not satisfactory. Ping An Financial’s account has released the BaaS platform. What is the situation?
Lu Yifan : There are some problems with the existing BaaS concept. We are introducing a new BaaS concept. We call it BNaaS. As for this N, it means that you can go to our recently published white paper .
There are two types of BaaS, one is that all nodes are in the cloud, which is no different from the database. And now most blockchain applications, to be honest, should use the database. Because there is no difference to the database for the customer. Another type of BaaS is supported for offsite deployment. This type of BaaS is also more like a blockchain.
Babbitt: It is said that the blockchain technology of the account book is not used by Ping An’s internal business. Why?
Lu Yifan : There are a lot of internal uses. It is possible to publicize the caliber problem. We use our technology because we can provide solutions for some very important problems that you can't solve, such as the 3D zero knowledge I just mentioned.
Babbitt: It is said that 壹通通 tries to land something around its financial business, but it does not produce much value. Is that right?
Lu Yifan : Of course it is not true. Although we are a financial technology company, the two words of finance are placed in front . Anything that cannot make money is a hooligan .
The difference between us and the outside blockchain team is that we have a KPI with balanced business revenue. And many well-known big projects are not necessarily cheap.
Babbitt: Some people say that Ping An is a big business, and it is easy to get a lot of bank projects. The company itself is an endorsement. So, do users believe in blockchain technology or believe in the brand of peace?
Lu Yifan : In fact, it should be reversed. Because it is a large enterprise, it will cause a small number of organizations to be jealous. Therefore, we use the blockchain as a weak centralization method, which can reduce the taboos of others , especially the full encryption framework. The data is mastered by ourselves, and the customers are also controlled by themselves, so that you will not let you It is taboo.
Babbitt: You used to say that traceability does not require blockchain technology, and traceability directly to the database + digital signature can be solved. So, are the blockchains on the market that are traceable to the source scenes all pseudo-blockchains? In addition to traceability, what other scenarios are not suitable for blockchain?
Lu Yifan : The deposit certificate is the least suitable for the blockchain . Because digital signatures are proof of tampering, most blockchain projects use digital signatures to solve them. To trace such things, you add timestamps to the database and do the same thing .
In general, in the following four value points, I have repeatedly stressed that the real advantage of not the blockchain:
First, it cannot be tampered with , and digital signatures solve this problem perfectly and efficiently 30 years ago. Second, traceable , any database can be time stamped, if you feel inappropriate, you can hash them. Third, to improve efficiency , the so-called efficiency improvement is the informationization of data. After data informationization, even if it is to the mobile APP, the automation efficiency of mobile APP will never be worse than the blockchain. Fourth, data sharing , there is absolutely no data sharing in the business world. In addition to poverty alleviation and public welfare projects, data sharing is a wishful thinking.
Babbitt: Some well-known people believe that the alliance chain does not bring any value to the blockchain technology. Blockchain disruptive innovation will not come from the alliance chain, but only from the public chain. Do you agree with this point of view?
Lu Yifan : The public chain can't escape the incentive mechanism. Once there is an incentive mechanism, it will oppose the current monetary policy, so the public chain can only develop in the entertainment field. What really brings industrial value is the alliance chain , and we also create value through the alliance chain.
Babbitt: In your opinion, is the blockchain a technology, a tool or a new production model? Because the alliance chain camp is discussing the real-time of the blockchain, it is often said that the use of blockchain technology will save the business how much cost, how much efficiency, and improve the customer experience. The public chain camp often says that the blockchain refactors the existing business model. For example, the incentive mechanism for introducing Token will be better than the existing company system in terms of benefit distribution.
Lu Yifan : The blockchain will definitely not save costs for the company, improve efficiency, and will not improve the customer experience. Because the point of improving efficiency is the informationization and digitization of the data itself, if the data is informatized and digitized, any technology can improve efficiency.
We do not need to introduce the Token mechanism. The incentive mechanism we introduce is business . After weak centralization, let them reduce the monopoly of large enterprises. Let's do business together. Generate business value through network effects .
The Token incentive mechanism will eventually return to a problem. This incentive mechanism becomes a unit of value exchange, that is, currency, which is issued by the central bank. If you endorse by assets, it becomes an asset. How are assets cashed?
Babbitt: In terms of technology and landing, what are the recent plans for Credit Link?
Lu Yifan : There are many, I will see it in the second half.
Babbitt: How do you evaluate the work you are doing?
Lu Yifan : We are doing something very meaningful.
Babbitt: As a representative of the regular army, what is the blockchain in your eyes? Where is its value and limitations? In the blockchain field, what are your respected experts?
Lu Yifan : Many people have also said that the blockchain is not a new technology, but the integration of existing technologies, such as the integration of cryptography, distributed systems, and distributed databases.
The most respected experts are scientists who have made outstanding contributions to cryptography, distributed systems and distributed databases. Although they may not directly engage in blockchains, what they do to science is very respectable.
Babbitt: In the field of blockchain, what are the problems in technology and application that need to be solved?
Lu Yifan : I said two years ago that the theoretical problem has not existed on FiMAX, and there will definitely be some problems in the implementation of the application.
Many people don't know how to find a solution, they just take a look at the problem, and there are security holes in the domestic blockchain . For example, hashing is a big security hole.
For example, in a famous blockchain application in China, the customer's address is hashed. Why is it hashed? Because I don't want to clear the text, but in fact the hash is equal to the clear text .
Because the hash can collide with the results, everyone in China has an address, a total of 1.4 billion addresses, on my laptop can complete the hash of China's 1.4 billion address within 1 hour and protect it The address collided.
In general, the participants of a coalition chain know the approximate scope and format of the hash data. As long as they know the approximate range, they have the ability to solve the plaintext behind the hash.
What is worrying is that this security risk is being output by many IT vendors as a solution, which will cause very bad feedback in the blockchain field.