Introduction: On April 23, at the UP2018 Tencent New Cultural and Creative Ecology Conference, Tencent released the first and currently only game "Let's Catch Monsters" that uses blockchain technology. Ma Huateng also stated at the recent 2018 China "Internet +" Digital Economy Summit that Tencent is using blockchain technology to cooperate with hospitals to issue digital prescriptions, and is considering promoting the implementation of this technology. In fact, compared with games and medical care, Tencent blockchain has natural application scenarios in finance. This article was written by abllen from the Billing Platform Department of the Tencent Technology Engineering Group. He introduced the latest application scenarios on the Tencent Cloud blockchain TBAAS system in detail from a technical point of view.
1. Introduction to Tencent Cloud Blockchain TBaaS
Tencent Cloud Blockchain TBaas is a blockchain platform launched in combination with cloud services. Its overall structure is as follows:
As a complete blockchain service, the features currently provided are:
(1) In operation, customers can quickly create a set of flexible and scalable self-owned applications according to their needs, including highly automated contract release, system monitoring and operation and maintenance.
(2) In terms of scheduling, the overall system achieves efficient scheduling and operation through kubernetes.
(3) In terms of security, strengthen user identity authentication based on CA and face.
(4) In terms of privacy protection, all data on the chain is encrypted, and can only be read and written after the relevant user or organization authorizes it.
(5) For contract editing, customers can self-write on the page IDE (see the figure below), and the back-end will do the relevant syntax and semantic legality checks and automatically submit it to the chain, so as to avoid contract security problems commonly encountered in the current industry.
6) On the underlying storage, almost unlimited expansion is achieved based on distributed storage.
7) In terms of performance, it has actually achieved 3000TPS (pbft), which is higher than the industry average and sufficient for general commercial applications.
8) In terms of data post-processing, TBaas also provides convenient services such as AI analysis.
In terms of specific security processing, TBaas fully supports the national secret algorithm and introduces CFCA (China Financial Certification Center) to meet the compliance requirements of domestic financial and other industry applications, realize legal traceability and arbitration judgment, and also support hardware encryption to further Improve the processing speed of encryption authentication. For contracts, based on the zero-knowledge proof method, all parties involved in the transaction can only process their own data, and at the same time realize the integrity of the overall transaction and cannot be tampered with.
The blockchain projects supported by TBaas include HyperledgerFabric and BCOS. At present, the underlying research and development of blockchain is mainly based on Fabric construction. Hyperledger is an open source project launched by the Linux Foundation in 2015 to promote blockchain digital technology and transaction verification. It has now evolved to version 1.1. Mainly used in alliance chain.
The main features of HyperledgerFabric are the generalized architecture and modular design, which makes it easy to adjust or optimize according to needs, and supports smart contract editing based on Go and Node.js, which is easy to use. In the Fabric architecture, three roles of Endorse Peer (endorsing node) / Commit Peer (all peers) / Orderer (consensus service) are clearly divided. The basic structure is as follows
In the whole process, the performance is improved as much as possible through the parallel execution of time-consuming operations such as initial endorsement and contract simulation execution, and then through consensus, the transaction sequence and verification are serialized according to time to achieve the consistency of ledger data, so as to achieve performance and consistency better balance.
In the specific development process, we have done the following exploration. The first is the overall system-level optimization, such as Signature offload, request signature processing under the TLS system is CPU intensive, and there are quite a few similar operations in the actual Fabric operation. By caching the repeated results, the processing capacity of a single chain can be effectively improved. The node's signature calculation and data landing, because the storage chain structure is serial in a single process, through the pipeline method, the signature calculation can be parallelized to greatly improve the overall performance. For system exceptions such as Kafka or the leveldb and Couchdb crash of the underlying storage, code-level modifications have also been made to ensure data loss.
The second is logic flow optimization. At present, Fabric adopts the optimistic locking method to simulate the execution and version number mechanism to ensure the consistency of chain data. However, for hotspot accounts and other situations with large conflicts, it is easy to lead to a high transaction failure rate. The improvements are: 1. Give up the version number and use incremental for read and write sets. 2. Use shadow accounts, that is, to split some hot accounts, and then periodically adjust the balance between these accounts.
In terms of core structure, in the current Fabric design, Orderer uses Kafka to achieve orderly requests. This is an effective design, and the actual test performance can reach 10,000 TPS. However, the centralization of Kafka itself leads to limited scalability and stability, and at the same time The management commands are also sent together through Kafka to ensure the consistent processing results of nodes, which may not take effect in time once messages accumulate. The solution, 1 pair of message queues, requires that management messages can be queued, that is, messages are prioritized; 2, a distributed consensus algorithm such as pbft is introduced. After applying pbft and implementing optimization, the measured data reaches 3000TPS. Customers can choose between two consensus methods according to actual scenario needs.
In terms of underlying storage, the fabric block structure is currently completely chained, which leads to the fact that even if there are two unrelated transactions, the verification and writing of nodes must be performed serially. The preliminary plan uses structures such as DAG or HashGraph to make operations parallel. to further improve performance.
The current blockchain applications, one type is various digital currencies, which are typical natural adaptation scenarios, building a specific ecology by trying incentives and other strategies, and the promotion and application in other industries, generally using the The characteristics of centralization and data immutability can solve the problems of mutual trust and information processing among multiple parties, such as supply chain financing and credit, commodity anti-counterfeiting, medical record and image sharing, IoT data security, cross-border clearing, real-time transaction reconciliation, etc. .
At this stage, the application of blockchain is still being explored and tried, and various practical landing cases are urgently needed to verify the feasibility and effectiveness of various technical solutions. To judge whether a scenario is suitable, we suggest to consider the following principles: 1) There are obvious consensus requirements 2) Multiple parties have the will to keep accounts 3) The data is open and multiple parties are equal (this does not mean that it can be viewed directly, it may be authorized to decrypt), Typically, such as transaction reconciliation, multiple parties have records, and they all need a clearly agreed result, which is more suitable.
At present, there are many TBaaS blockchain implementation scenarios. Here are two typical cases: one is Love Life, which focuses on insurance claims, and the other is Weite, which focuses on supply chain finance.
In the blockchain cooperation with Love Life, it mainly explores the organization of medical institutions, insurance companies, health information platforms, etc. into a blockchain alliance, opening up many related links, and storing digital certificate information on the blockchain safely and efficiently. Strive to fundamentally solve the application and technical problems such as the security and relevance of medical digital information, realize the safe sharing and interconnection of medical and insurance information in the true sense, and provide users with efficient direct access, safe and affordable health care and insurance protection Serve.
The specific method is as shown in the figure. Institutions are divided into medical institutions, insurance companies, case centers, and supervision nodes. Here, all parties, such as medical institutions and insurance companies, are actually composed of multiple organizations. To simplify, only one is listed in the figure.
In actual insurance claims, the trouble for customers lies in the collection and arrangement of case documents, and the difficulty for insurance companies lies in the screening and validity verification of cases. These are two parties that are difficult to trust, and there is a certain cost to pay (such as cross-border insurance). Hospital processing), and the hospital, as an intermediary, bears the responsibility for the custody of the cases, and in reality also needs to check the cases of other medical institutions. This satisfies several requirements for blockchain applications at the same time: the need for consensus between customers and insurance companies, the need for multi-party records between hospitals, and the need for open data sharing. Considering the inconvenience of customer bookkeeping, case centers and supervisory nodes (optional) are introduced to participate in third-party bookkeeping, which together constitute a typical blockchain case record scenario.
Because the block data cannot be tampered with, insurance companies can obtain detailed case documents to achieve effective and accurate claims as long as they are authorized by the user and approved by the medical party. For insurance companies, it only needs to pay a small cost, and it is more convenient for users to have no cost. The cost paid by insurance companies will be transformed into material incentives for medical institutions and case centers, further promoting the healthy operation of the entire blockchain.
When the costs paid by insurance companies are transferred to multi-party medical institutions and case centers, there is a need for regular reconciliation and settlement. This is the typical reconciliation scenario mentioned above. The transaction information when the actual medical institution agrees (that is, endorses) On the chain, through smart contracts, all parties can clearly determine their own income locally, and achieve accurate and decentralized reconciliation. That is to say, through the blockchain, the problem of data verification and revenue sharing between all parties in the insurance claim process can be solved.
Supply chain finance is another relatively suitable adaptation scenario for blockchain. Based on the decentralized sharing, trust and security mechanism of blockchain, it can effectively solve the transfer, transaction, and supervision issues related to supply chain finance. The micro-specialization scenario mainly digitizes various supply chain tangible assets, such as agricultural products, which can be pre-purchased in the form of digital bills (to allow more people to participate through the blockchain) half a year in advance, and the delivery will be due after half a year. In case of shortage of funds during this period, the pre-purchaser can sell some of the bills in advance, and when they expire, they can transfer the tickets to wholesalers or end users, and then the wholesalers can directly pick up the goods. This can save the cost of warehousing and delivery, reduce intermediate steps, and reduce the ineffective cost of information asymmetry, and improve the transaction efficiency of the entire agricultural product through new Internet technologies.
At present, the blockchain is still an emerging field. There are constantly new and exciting new technologies proposed. Some data mention that the blockchain has evolved to 4.0. In terms of the industry, it is generally demand that drives technological progress, or the choice is made by demand when the surrounding technologies are basically mature and paved, while the current blockchain is more based on the characteristics of technology to extend or Building a demand, not an inevitable demand, is a key issue that needs to be solved urgently in the current blockchain development. In this article, two practical examples are presented for your reference.
With the development of the industry, people and systems and systems (such as various IOT devices) are increasingly connected, and the traditional centralized architecture will inevitably face challenges in terms of scalability, while the decentralized blockchain should be Alternative solutions with good prospects. The billing platform department will combine its accumulation in distributed system transactions and algorithms to continuously solve the technical problems faced by the blockchain. I believe that with the continuous evolution of the blockchain, there will be more application scenarios to show its value.