Write Request Throttling via Completion Time Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The mismatch in access speeds between cache memory and hard disk devices in storage systems can lead to cache overflow, resulting in delayed write requests and potential system failures due to inefficient write throttling mechanisms.
Innovation Solution
A method is introduced to determine a predetermined completion time for write requests based on the allocated storage space and release rate from a high-speed storage device to a low-speed device, triggering a throttling indication when the completion time exceeds a threshold, thereby proactively controlling the issuance of further write requests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If data is buffered in cache memory to provide quick response to write requests, then write response speed is improved, but cache overflow occurs due to mismatch between cache access speed and hard disk access speed
Solution Approach 1:
The patent calculates the completion time of write requests in advance by determining the storage space occupied by pending write requests and dividing it by the data release rate from cache to hard disk. This preliminary calculation allows the system to predict potential cache overflow before it occurs and trigger throttling proactively, rather than reactively responding to actual overflow conditions.
Solution Approach 2:
The patent implements a feedback mechanism where the storage controller monitors the completion time of write requests and compares it against a threshold. When the completion time exceeds the threshold, the system sends a throttling indication to the host, which then reduces the write request rate. This closed-loop feedback control prevents cache overflow while maintaining optimal write performance.
2Reliability
If write requests are delayed in cache memory due to hard disk access speed limitations, then data integrity is maintained, but system failure is triggered due to excessive delay time
Solution Approach 1:
The system proactively calculates the completion time of write requests before they actually complete, using the formula: completion time = storage space occupied by pending write requests / data release rate. By comparing this predicted completion time against a threshold in advance, the system can trigger throttling before excessive delays occur, preventing system failure while maintaining data integrity.
Solution Approach 2:
The patent applies preliminary anti-action by triggering write throttling before the cache actually overflows or delays become excessive. The system predicts potential problems by calculating completion times and takes preventive action by sending throttling indications to the host, thereby counteracting the potential harmful effect of excessive delay before it occurs.
3Reliability
If the storage controller monitors write request completion actively, then system reliability is improved, but computational overhead increases
Solution Approach 1:
The storage controller performs self-service by autonomously calculating the completion time of write requests using internally available data (storage space occupied by pending requests and data release rate). This self-calculated completion time is then used to trigger throttling without requiring complex external monitoring systems or additional hardware, reducing overall system complexity while maintaining reliability.
Data Source
AI summary
In certain techniques, a size of a storage space in a first storage device that is allocated for write requests is determined. Write requests are issued by a control device of the storage system to store data into the first storage device and are waiting to be completed. A predetermined release rate of releasing the data stored in the first storage device to a second storage device is determined. An access speed of the first storage device is greater than that of the second storage device. A predetermined completion time of the write requests is determined based on the size of the storage space and the predetermined release rate, and in response to the predetermined completion time exceeding a predetermined time threshold, a throttling indication is transmitted to the control device such that the control device throttles issuing of a further write request for the first storage device.