Event Routing in Distributed Data Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distributed computing systems with distributed data management, the current event routing methods lead to increased communication burden and resource wastage due to the unnecessary transmission of events to server processes that are not associated with interested application processes, resulting in higher latency and reduced throughput.
Innovation Solution
Optimized event routing is achieved by registering the interests of application processes with both primary and backup server processes, allowing events to be communicated only to server processes associated with interested application processes, thereby reducing redundant calculations and communications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If events are communicated to all server processes serving application processes, then all application processes can receive events, but communication burden and resource wastage increase due to unnecessary transmissions to server processes without interested applications
Solution Approach 1:
The patent applies preliminary action by pre-registering application process interests with server processes before events occur. Each server process maintains a registry of which application processes are interested in which events, allowing it to filter events locally without needing to broadcast to all servers. This advance preparation enables efficient event routing when events actually occur.
Solution Approach 2:
The patent introduces an intermediary mechanism where server processes act as mediators between the data fabric and application processes. Server processes receive events from the data fabric, filter them based on registered interests, and forward only relevant events to associated application processes. This intermediary layer eliminates the need for direct communication between all server processes and all application processes.
2Loss of energy
If events are filtered at each server process to determine application interest, then unnecessary communications are reduced, but computational burden increases due to repeated interest calculations
Solution Approach 1:
The patent resolves this contradiction by performing the filtering computation in advance during event registration. When application processes register their interests, server processes pre-compute and store which applications are interested in which events. This eliminates the need for repeated filtering computations when events occur, as the filtering logic is already determined and stored.
Solution Approach 2:
The patent uses copying by maintaining copies of interest registration information at server processes. Each server process stores a local copy of which application processes are interested in which events, allowing it to independently filter events without needing to query application processes or perform repeated calculations. This copied information enables fast, computation-free filtering at event time.
3Reliability
If the data fabric communicates events to all server processes, then no events are lost, but system latency increases due to redundant transmissions
Solution Approach 1:
The patent applies taking out by extracting the filtering function from the data fabric and moving it to server processes. Instead of the data fabric broadcasting events to all server processes, the data fabric only sends events to server processes that have registered interests. This extraction of the filtering function eliminates redundant transmissions while ensuring all interested applications receive events, reducing latency without sacrificing reliability.
Solution Approach 2:
The patent reduces latency by performing preliminary registration of application interests with server processes before events occur. This advance registration allows server processes to immediately filter and forward relevant events without delay, eliminating the need for post-event filtering and reducing overall event distribution latency.
Data Source
AI summary
Event routing in a distributed computing system with distributed data management is optimized by communicating events a datum only to either backup server processes or server processes respectively associated with an application process having an interest in an event affecting the datum.