Round-Based Resource Scheduler With Latency Offset
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing scheduling methods for shared resources do not effectively manage unscheduled usage requests, leading to inefficiencies and potential latency issues, particularly when resource users have varying bandwidth and latency requirements.
Innovation Solution
A method and scheduler that allocate a per-round usage allowance to each resource user, scheduling unscheduled requests within the current round up to the allowance and offsetting excess requests to later rounds based on a specified maximum request latency, ensuring both fair access and meeting quality-of-service requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If round robin scheduling is used to allocate resource access equally among users, then fairness is improved, but latency requirements may not be met when users have varying bandwidth needs
Solution Approach 1:
The scheduling is segmented into discrete rounds, with each round allocating a specific usage allowance to each user. This segmentation allows the system to track and enforce both fairness (through equal round-based allocation) and latency requirements (by calculating round offsets for excess requests), resolving the contradiction between equal treatment and differentiated service levels.
Solution Approach 2:
The system dynamically adjusts the treatment of different users based on their remaining unscheduled requests and latency requirements. While the base allocation remains fair and equal per round, the system dynamically calculates round offsets for users with stricter latency requirements, allowing them to have their excess requests scheduled earlier in subsequent rounds, thus adapting to varying bandwidth and latency needs.
2Productivity
If per-round usage allowance is allocated to each user, then bandwidth control is improved, but complexity of scheduling increases
Solution Approach 1:
The system performs preliminary actions by pre-calculating the number of round offsets needed for each user's excess requests based on their latency requirements. This preliminary calculation is done before final scheduling, allowing the scheduler to efficiently allocate bandwidth while managing complexity through structured, pre-determined offset values rather than complex real-time calculations.
Solution Approach 2:
The scheduling mechanism incorporates feedback loops where the scheduler continuously monitors each user's allocated versus actual usage, tracks remaining unscheduled requests, and adjusts subsequent round allocations accordingly. This feedback-based approach enables efficient bandwidth control while maintaining manageable complexity through iterative, rule-based adjustments rather than complex optimization algorithms.
3Ease of operation
If excess requests are scheduled for later rounds, then fairness is maintained, but latency may increase beyond acceptable limits
Solution Approach 1:
The system applies local quality by treating different users differently based on their specific latency requirements. While the overall system maintains fairness through equal per-round allocation, individual users with stricter latency requirements receive localized adjustments in the form of calculated round offsets, allowing their excess requests to be prioritized in specific future rounds without compromising the overall fairness of the scheduling system.
Data Source
AI summary
Methods of scheduling usage of a resource between multiple resource users and corresponding schedulers are disclosed. The scheduling is performed by initially allocating a per-round usage allowance to each resource user of the multiple resource users and then determining a usage schedule for the resource in a sequence of rounds. At each round the determining comprises, for each resource user, when there are unscheduled usage requests for the resource user, scheduling a portion of the unscheduled usage requests which does not exceed the per-round usage allowance for the resource user, wherein the portion is scheduled for resource usage in association with the current round. Further, at each round, when there exists a remainder of the unscheduled usage requests which exceeds the per-round usage allowance for the resource user, the remainder is scheduled in association with a later round, wherein a number of rounds offset from the current round to the later round is calculated in dependence on a specified maximum request latency for the resource user. Minimum bandwidth and maximum latency requirements for each user are thus supported.