Parallel Log Entry Duplication in Blockchain Node Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing data recording process in node clusters, where nodes operate in follower, candidate, and leader states, suffers from low efficiency due to serial processing of log entries, leading to delays and potential network issues like missing or over-receiving log entries.
Innovation Solution
Implementing a method where a leader node sends multiple log entries in parallel to follower nodes, with confirmation responses and supplementary log entries sent accordingly, using multithreading and receive windows to ensure data consistency and efficiency, and maintaining a confirmation table to manage commit indices and packet loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If log entries are sent serially to follower nodes, then data consistency is maintained, but transmission efficiency is low
Solution Approach 1:
The patent segments the log entry transmission process by introducing receive windows for different log entry types (first receive window for committed log entries, second receive window for uncommitted log entries). This allows parallel processing of different log entry segments while maintaining consistency through selective application based on commit indices.
Solution Approach 2:
The patent implements dynamic transmission by allowing the leader node to send multiple log entries in parallel based on the follower node's current commit index. The system dynamically adjusts which log entries to send next based on real-time consensus state, transforming the static serial process into a dynamic parallel one.
2Productivity
If multiple log entries are sent in parallel, then transmission efficiency is improved, but network reliability decreases due to potential packet loss or duplicate reception
Solution Approach 1:
The patent implements feedback mechanisms where follower nodes send confirmation responses to the leader node after receiving and processing log entries. The leader node uses this feedback to track which log entries have been successfully committed and adjusts subsequent transmissions accordingly, ensuring reliability even with parallel sending.
Solution Approach 2:
The patent uses preliminary action by maintaining a commit index that records which log entries have been successfully committed before sending subsequent log entries. This preliminary tracking prevents duplicate processing and ensures that only valid log entries are applied, maintaining data consistency despite parallel transmission.
3Reliability
If the leader node waits for confirmation responses before sending next log entries, then data consistency is ensured, but processing speed decreases
Solution Approach 1:
The patent applies partial action by sending only the necessary portion of log entries that can be safely processed based on the current commit index. Instead of waiting for complete confirmation of all log entries, the system sends uncommitted log entries through a separate channel (second receive window) that can be processed asynchronously, achieving partial progress without compromising consistency.
Solution Approach 2:
The patent segments log entries into committed and uncommitted categories with separate transmission channels. Committed log entries use the first receive window with strict confirmation requirements, while uncommitted log entries use the second receive window with more flexible processing, allowing different speeds for different consistency requirements.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
The embodiments of the present invention disclose a log entry duplication method and apparatus, a computer device and a storage medium, and belong to the technical field of block chains. The first node sends in parallel a plurality of first log entries to a plurality of second nodes, and provides a supplementary sending mechanism for the log entries, so as to avoid the situation that the second nodes receive log entries by mistake such as missing receiving or over-receiving, guarantee the consistency of data of each node, save the time for sending the plurality of the first log entries, and improve the log entry duplicating efficiency.