Low Latency Messaging via Priority-Based Direct Injection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-tenant database systems face challenges in efficiently processing and transmitting messages with low latency, leading to increased resource overhead and user dissatisfaction due to batch-wise processing methods that result in significant delays.
Innovation Solution
Implementing a direct injection and tickle mechanism for triggered send requests, where high-priority messages are directly processed by selecting an available message server for immediate processing, and medium or low-priority messages are processed using a tickle notification to wake up idle message servers, reducing latency and improving message delivery efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If batch-wise periodic processing is used for message transmission, then resource overhead is reduced and system stability is maintained, but message transmission latency increases significantly
Solution Approach 1:
The system dynamically adjusts message processing mode based on priority level. High-priority messages trigger immediate processing through direct injection or tickle mechanisms, while medium/low-priority messages use batch processing. This dynamic adaptation resolves the contradiction by allowing the system to be responsive when needed while maintaining efficiency during normal operation.
Solution Approach 2:
The message queue is segmented into multiple priority levels (high, medium, low). Each segment is processed differently: high-priority messages receive immediate attention through direct injection or tickle notifications, while other segments continue with batch processing. This segmentation allows the system to reduce latency for critical messages without compromising overall processing efficiency.
2Loss of time
If direct injection mechanism is used for high-priority messages, then message transmission latency is reduced, but system complexity increases
Solution Approach 1:
The system applies different processing qualities to different message segments based on their priority. High-priority messages receive the premium direct injection treatment with immediate processing, while medium and low-priority messages use the simpler batch processing approach. This local differentiation reduces overall system complexity while still achieving low latency for critical messages.
Solution Approach 2:
Instead of implementing direct injection for all messages (which would maximize complexity), the system applies this action only partially to high-priority messages where it is most needed. This partial application achieves the latency reduction benefit while limiting the increase in system complexity to only what is necessary for critical operations.
3Speed
If tickle notification is used to wake idle message servers, then message processing responsiveness is improved, but resource consumption increases
Solution Approach 1:
The tickle notification mechanism is applied selectively only to medium-priority messages, not to all messages. High-priority messages use direct injection, while low-priority messages remain in batch processing. This partial application of tickle notifications improves processing speed for important messages while limiting resource consumption increases to only what is necessary for medium-priority operations.
Data Source
AI summary
Systems, methods, and computer-readable media for processing and sending messages with low latency are described. An application server may obtain a triggered send request (TSR) message from triggered send (TS) subscriber. The TSR message may comprise TSR information and a TSR payload. The application server may directly inject the TSR payload to an available message server for processing when the TSR information indicates that a first priority is associated with the payload, and may notify the available message server that the TSR payload is available for processing when the TSR information indicates that a second priority is associated with the payload. The application server may also write the TSR payload and/or TSR information to a TSS queue of a TSS tenant space, and may write the payload to a non-relational datastore regardless of a priority associated with the payload. Other embodiments may be described and/or claimed.