Unicast Cluster Messaging via Group Leader Intermediary
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing application server network clusters face challenges in scalable and reliable messaging, particularly in scenarios requiring broadcast of messages to multiple servers without the limitations of multicast sockets, such as liveliness detection and session replication.
Innovation Solution
Implementing a unicast-based cluster messaging system where messages are sent through a group leader to other group leaders and members, with heartbeats for leader monitoring and retransmission to ensure delivery, allowing for flexible configuration between unicast and multicast modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If unicast messaging is used instead of multicast sockets, then reliability of message delivery is improved, but the number of message transmissions increases
Solution Approach 1:
The patent introduces a group leader as an intermediary node in each cluster group. Instead of every server sending unicast messages to every other server (N transmissions), each server sends messages only to its group leader (1 transmission per server). The group leader then distributes the message to other group members. This reduces the number of transmissions from O(N^2) to O(N) while maintaining unicast reliability.
Solution Approach 2:
The patent divides the cluster into multiple groups, each with a designated group leader. This segmentation allows the messaging system to handle large clusters by breaking them into smaller manageable units. Each group independently manages its message distribution, reducing the overall communication overhead while maintaining reliable delivery across the entire cluster.
2Reliability
If multiple unicast messages are sent for broadcast, then message delivery to all members is ensured, but network traffic increases
Solution Approach 1:
The group leader acts as a mediator that consolidates broadcast messages. Instead of each server sending separate unicast messages to every other server (creating N*(N-1) traffic), each server sends one message to its group leader, and the leader distributes it to other group members. This reduces network traffic from O(N^2) to O(N) while ensuring reliable delivery to all members.
Solution Approach 2:
The patent merges multiple unicast transmissions into a single broadcast from the group leader. When a server needs to communicate with multiple members, the message is first sent to the group leader, which then combines the distribution task and sends to all group members efficiently. This merging reduces redundant network traffic while maintaining delivery reliability.
3Productivity
If cluster size increases, then scalability is improved, but message dissemination complexity increases
Solution Approach 1:
The patent segments the cluster into multiple groups with designated leaders. This allows the system to scale to large numbers of servers by organizing them into smaller groups. Each group independently manages its own message distribution, so adding more servers only adds more groups or members to existing groups, not complexity to the entire system. The complexity per group remains constant regardless of total cluster size.
Solution Approach 2:
The patent introduces a hierarchical dimension to message dissemination: servers first communicate within their group (local dimension), and group leaders communicate with other leaders (global dimension). This two-level hierarchy allows the system to scale horizontally by adding more groups, while each group maintains manageable complexity. The problem transforms from a single flat N×N communication matrix to a hierarchical structure with controlled complexity at each level.
Data Source
AI summary
A system for clustering messaging in application server specific environment can comprise group leader application server nodes that are inter-connected to each other directly and with group member application server nodes. A group can include at least one group leader application server node and at least one group member server nodes. The group member application server nodes can send and receive messages through associated group leader application server nodes.