Fractional Queue Depth Management via Token Bucket
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional queue depth management in information handling systems is limited to integer increments, making it difficult to adjust queue depth in small percentage increments, which can lead to inefficiencies in processing I/O requests, especially in systems with a large volume of requests.
Innovation Solution
Implementing a token bucket scheme that allows for fractional queue depth management by adding virtual tokens to a virtual bucket based on a mathematical combination of the queue depth limit and latency of processed I/O requests, enabling more precise control over the number of I/O requests processed over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional integer-based queue depth management is used, then the system is simple to implement, but the ability to adjust queue depth in small increments is lost
Solution Approach 1:
The patent segments the queue depth limit into integer and fractional components. The integer portion (QLIM_INT) is managed by traditional integer-based mechanisms, while the fractional portion (QLIM_FRAC) is managed by a token bucket scheme. This segmentation allows precise control over queue depth adjustments in small increments while maintaining simplicity in the overall system architecture.
Solution Approach 2:
The patent introduces a token bucket as an intermediary mechanism between the integer queue depth limit and the actual I/O request processing. The token bucket contains fractional tokens that act as a mediator, allowing smooth, incremental adjustments to queue depth without requiring changes to the underlying integer-based queue management infrastructure.
2Productivity
If queue depth threshold is increased by integer increments, then the management scheme is simple, but the efficiency of processing I/O requests decreases
Solution Approach 1:
The patent makes the queue depth limit dynamic by introducing a token bucket that can be continuously refilled with fractional tokens. Instead of static integer increments, the system dynamically adjusts queue depth by adding fractional tokens at rates proportional to processing capacity, allowing the queue depth to adapt smoothly to changing system conditions and maximize I/O processing efficiency.
Solution Approach 2:
The patent changes the parameter representation from discrete integers to continuous fractional values. By representing queue depth as QLIM = QLIM_INT + QLIM_FRAC, where QLIM_FRAC is controlled by the token bucket, the system can make fine-grained parameter adjustments that directly improve I/O processing efficiency without sacrificing operational simplicity.
3Measurement precision
If fractional queue depth management is implemented, then the control precision over I/O requests is improved, but the system complexity increases
Solution Approach 1:
The patent segments the queue depth management into two independent parts: integer-based queue limit management and fractional-based token bucket management. This segmentation isolates the complexity of fractional control into a dedicated token bucket mechanism, allowing precise queue depth control while keeping the overall system architecture understandable and maintainable.
Solution Approach 2:
The token bucket serves as an intermediary layer that handles all fractional queue depth control operations. By confining fractional management to this dedicated intermediary mechanism, the patent achieves high control precision without propagating complexity throughout the entire queue management system. The integer queue limit and fractional token bucket work together through well-defined interfaces.
Data Source
AI summary
A method for managing input/output (I/O) traffic in an information handling system. The method may include receiving electronic I/O requests from a network-attached server, determining a queue depth limit, monitoring latency of processed electronic I/O requests, and processing received electronic I/O requests. The number of electronic I/O requests permitted to be processed over a period of time may be based on a mathematical combination of the queue depth limit and a latency of processed electronic I/O requests. The determined queue depth limit may be a fractional value.