Memory Subsystem Message Retransmission via Command Queue Depth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In memory subsystems, the retransmission of messages based on command queue depth is challenging due to the risk of causing interrupt storms and performance delays, as existing methods do not effectively consider the host system's processing load and timing for response messages.
Innovation Solution
The memory subsystem retransmits messages upon detection of an indication of the processing of a number of commands from the host system, using either the expiration of a time period or a count of commands reaching a threshold count equal to the command queue depth as an indication, to avoid overwhelming the host system and ensure timely response receipt.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the memory subsystem retransmits messages frequently to ensure timely receipt, then message delivery reliability is improved, but interrupt storms occur and host system performance deteriorates
Solution Approach 1:
The memory subsystem uses periodic retransmission based on command queue depth thresholds. Instead of continuous or frequent retransmission, the system waits for the command queue depth to reach a threshold (e.g., 50% or 80% full) before retransmitting messages, creating a periodic action that balances reliability with host system performance
Solution Approach 2:
The retransmission strategy dynamically adapts to host system conditions by monitoring command queue depth. The system adjusts its retransmission behavior based on real-time queue depth measurements, increasing retransmission activity when the host is idle (queue depth low) and reducing it when the host is busy (queue depth high), thereby preventing interrupt storms while maintaining message delivery reliability
2Object-affected harmful factors
If the memory subsystem waits for host processing before retransmitting, then interrupt storms are avoided, but message delivery time increases
Solution Approach 1:
The memory subsystem performs preliminary monitoring of command queue depth and proactively retransmits messages when the queue depth reaches a threshold, before the host system becomes overloaded. This preliminary action prevents interrupt storms by avoiding retransmission during high-load periods, while still maintaining relatively timely message delivery through early retransmission opportunities
Solution Approach 2:
The system uses feedback from command queue depth monitoring to control retransmission timing. By continuously monitoring the queue depth and using it as a feedback signal, the memory subsystem intelligently determines when to retransmit messages, balancing the need for timely delivery with the need to prevent interrupt storms based on actual host system state
Data Source
AI summary
A method and system for retransmitting messages in a memory subsystem are described. A message is transmitted to a host system. A response message is expected to be received from the host system in response to the message. A determination that the response message was not received prior to detecting an indication of a processing of a number of commands from the host system is performed. The message is retransmitted to the host system in response to the determination.