CN-122022809-A - High throughput block chain partition management method and system

CN122022809ACN 122022809 ACN122022809 ACN 122022809ACN-122022809-A

Abstract

The invention belongs to the technical field of blockchains, and particularly relates to a high-throughput blockchain partition management method and system. The method comprises the steps of S1, receiving a transaction submitted by a client, judging whether the transaction is a single-pool transaction or a cross-pool transaction according to a mapping relation between an account related to the transaction and a transaction pool, executing step S2 if the transaction is the single-pool transaction, executing step S3 if the transaction is the cross-pool transaction, independently completing local consensus and block generation and account book updating of the transaction by each transaction pool based on an on-chip consensus protocol, S3, executing a global coordination flow if the transaction is the cross-pool transaction, and executing the global coordination flow as follows, wherein each transaction pool submits a block which is completed by the local consensus of the cross-pool transaction set to the global cache pool, and then uniformly coordinates and executes batch consensus blocks by the global cache pool to transfer the executed cross-pool transaction batch to a target transaction pool.

Inventors

SHEN JIAN
MA HAOYU
SHENG XINLEI
WANG CHEN
ZHOU TIANQI
TAN HAOWEN
YANG HUIJIE
LIU JIQIANG
HUANG WEI
XIE LEI

Assignees

浙江理工大学
现代纺织技术创新中心(鉴湖实验室)

Dates

Publication Date: 20260512
Application Date: 20260410

Claims (7)

1. The high throughput block chain fragment management method is characterized by comprising the following steps: S1, transaction receiving and type judging: The system receives the transaction submitted by the client and judges that the transaction is a single-pool transaction or a cross-pool transaction according to the mapping relation between the account related to the transaction and the transaction pool, if the transaction is the single-pool transaction, the step S2 is executed, and if the transaction is the cross-pool transaction, the step S3 is executed; S2, single-pool transaction processing: if the transaction is a single-pool transaction, independently completing local consensus, block generation and account book update of the transaction by each transaction pool based on an on-chip consensus protocol; S3, cross-pool transaction processing: if the transaction is a cross-pool transaction, executing a global coordination flow, wherein the global coordination flow is as follows: and uniformly coordinating and executing batch consensus blocks by the global cache pool, and forwarding the executed cross-pool transaction batch to the target transaction pool.
2. The high throughput blockchain segment management method of claim 1, wherein in step S1, all accounts involved in the single-pool transaction fingers belong to the same transaction pool, and accounts involved in the cross-pool transaction fingers belong to at least two different transaction pools.
3. The high throughput blockchain partition management method of claim 1, wherein step S2 includes the steps of: s21, under the single transaction pool scene, after receiving single pool transactions, a leading node in the current pool performs voting according to a consensus protocol, and a duplicate node participates in voting and confirmation in parallel; s22, after the local consensus of each transaction pool is completed, the block is directly written into a local account book, and the corresponding account state is updated without cross-pool coordination.
4. The high throughput blockchain partition management method of claim 1, wherein in step S3, the global cache pool exists as a buffer table in a local memory of each replica node, each replica node maintains a buffer status view, and status consistency is ensured by a periodic consensus protocol.
5. The method of claim 4, wherein in step S3, the global buffer pool operates with a fixed time window delta as a scheduling period, and each period is summarized with a plurality of cross-pool transaction blocks from different transaction pools, and at the end of the time window, a batch of transaction blocks is constructed by a current round leader node of the global buffer pool And initiates all copies in one slice to participate in the consensus process.
6. The high throughput blockchain partition management method of claim 5, wherein step S3 includes the steps of: S31, cross-pool transaction submission: Source transaction Pool i local consensus generation block B i generates a cross-Pool transaction set according to cross-Pool transaction screening rules Pool i leader submits the collection to global cache Pool GBP, with local certificate QC i ; s32, global cache pool batch processing: Setting a fixed time window delta, broadcasting and triggering a window starting time t start by a GBP leader node, and collecting a cross-pool transaction set in the time window delta by a global cache pool Recording a global transaction list to form an ordered set, constructing a global block B G based on the transaction mapping of the target pool group, wherein the global block B G comprises all T global ; Broadcasting information to the copy nodes to trigger a global cache pool consensus process to generate global certificates Recording the blocks and pushing the blocks to each related transaction pool in parallel; s33, cross-pool transaction distribution and execution: Each transaction Pool analysis B G extracts a transaction tx j belonging to Pool j according to a target transaction Pool dst_pool, constructs a local block B j , adds a global certificate sigma G as a certificate, broadcasts and submits the certificate to a duplicate node, the duplicate node verifies sigma G validity, confirms that the transaction is globally approved, executes a consensus process, and generates a certificate QC j ;Pool j of a block B j as a target Pool for receiving and executing a corresponding cross-Pool transaction; Finally, the target account balance is updated, and tx j is marked as submitted.
7. A high throughput blockchain slice management system for implementing the high throughput blockchain slice management method of any of claims 1-6, the high throughput blockchain slice management system comprising: the transaction receiving module is used for verifying the transaction signature and account mapping and judging the transaction type; The local consensus module is used for executing a consensus protocol and generating a local block and a certificate; the global cache coordination module is used for maintaining a cache state table and running global batch consensus; The state updating module is used for verifying the validity of the global certificate and simultaneously executing account book updating and transaction submitting; and the synchronization and cleaning module is used for periodically synchronizing the cache state and cleaning the expired transaction records.

Description

High throughput block chain partition management method and system Technical Field The invention belongs to the technical field of blockchains, and particularly relates to a high-throughput blockchain partition management method and system. Background With the continuous expansion of the application ecology of the blockchain, the performance bottleneck of the underlying architecture is increasingly prominent. Traditional single-chain models are limited by modes of full-network node full-volume processing transactions, and the overall throughput of the system is limited by structural upper limits, so that horizontal expansion capability is difficult to obtain. In the face of the increasing high concurrency load, there is a need to construct a high performance blockchain system with linear expansion capability under the premise of maintaining decentralization and safety, and the direction has become an issue of common attention in academia and industry. The partition technology is proposed and applied as a horizontal extension paradigm of a blockchain, effectively solves the problem of single-chain performance constraint, and has the basic idea that a node set and a state space are divided into a plurality of parallel subdomains in a limited way, each subdomain is used as a partition to independently bear transaction receiving, executing and consensus in the scope of the partition, and a local account book and a state root are maintained. Inside the slice, to maintain state consistency and fault tolerance, a bayer fault tolerance type consensus protocol (Byzantine Fault Tolerance, abbreviated BFT) is typically used as a basis. The existing block chain system still faces structural bottlenecks in the execution of high-concurrency transaction, and is mainly embodied in the aspects of serial proposal flow, complex state consistency maintenance process, tight coupling of execution models and the like. Most of current protocols adopt a single leader node to drive a consensus flow, and the transaction needs to be sequentially processed by the single leader node and a broadcast proposal, which becomes a bottleneck of system performance. Secondly, in transaction coordination across fragments, a common two-stage commit mechanism requires locking of related fragments to ensure atomicity, which is prone to causing conflicts and causing resource blocking, and significantly reduces system performance. In addition, the transaction execution flow is highly coupled with the consensus submitting process, and a unified intermediate state or caching mechanism is lacked, so that each transaction needs to repeatedly access the global state in a plurality of stages, and the state tracking burden is heavy and the response delay is high. These problems are further amplified in the high frequency cross-slice transaction invocation scenario, and no solution currently achieves structural optimization at the three levels of communication, execution and data consistency. The document with the patent grant publication number CN113568974B of China provides a slice consensus method, and transaction processing is realized through local consensus and whole network sequencing in slices. However, in a large-scale high concurrency scene, the scheme still has the defects that cross-slice transactions are required to be split for execution and rely on whole network sequencing, the coordination expense is large, the intra-slice processing mode is single, the whole concurrency of the system is limited, and when the cross-slice call is frequent, the problems of conflict, delay and the like are easy to occur in the transaction scheduling and the state maintenance. Therefore, it is difficult to meet the performance and efficiency requirements in a high concurrency environment in the prior art. For a high concurrency scenario, the transaction is sequentially processed by the single leader node, so that the length of a transaction processing queue is obviously increased, and the throughput rate of the system is further reduced. The current framework cannot fully utilize the multi-node parallel processing capability in the slices, and all transactions need to be strictly and orderly executed in a single flow, so that the resource utilization rate is low and the response is delayed. Especially in frequently interacted application scenarios, the heavy synchronization overhead and message passing delay of the transaction confirmation process further limit the system performance. Existing schemes lack an efficient cross-stage parallel pipeline or intermediate cache mechanism to alleviate dependencies and latency between different stages. Therefore, it is very important to design and provide a blockchain consensus and execution method capable of improving the parallel processing capacity in the slices, reducing the cross-pool transaction coordination overhead and shortening the transaction confirmation delay on the premise of ensuring the system safety and con