Dynamic Batch Size Adjustment for Wide Area Network Prefetching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In wide area networks, determining the optimal batch size for data prefetching is challenging as prefetching too little data may result in lost opportunities for optimization, while prefetching too much data can increase client-perceived latency instead of improving it.
Innovation Solution
A device and method that measure the prefetch transfer time of a previous batch of data and compare it to a threshold value to determine and modify the size of the next batch for optimal prefetching, ensuring the prefetch transfer time approaches a target level.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the batch size for data prefetching is increased, then bandwidth utilization is improved, but client-perceived latency increases
Solution Approach 1:
The patent implements dynamic adjustment of prefetch batch sizes based on observed network conditions and transfer times. The system monitors actual prefetch performance and adapts the batch size parameter in real-time, transitioning from static to dynamic control to optimize the trade-off between bandwidth utilization and latency.
Solution Approach 2:
The system incorporates feedback mechanisms by measuring prefetch transfer times and using these measurements to adjust subsequent batch sizes. This closed-loop control ensures that the system learns from past performance and continuously optimizes the batch size to balance bandwidth efficiency with latency requirements.
2Loss of time
If the batch size for data prefetching is decreased, then client-perceived latency is reduced, but bandwidth utilization deteriorates
Solution Approach 1:
The system dynamically adjusts batch sizes based on network conditions rather than using fixed small batches. This allows the system to increase batch sizes when network conditions permit, thereby improving bandwidth utilization without permanently compromising latency performance.
Solution Approach 2:
The patent changes the batch size parameter adaptively based on observed transfer times and network conditions. By modifying this key parameter dynamically, the system can optimize bandwidth utilization when conditions allow while maintaining acceptable latency levels.
3Loss of time
If prefetching is implemented without optimization, then perceived latency is reduced, but network costs increase
Solution Approach 1:
The system uses feedback from measured prefetch transfer times to optimize batch sizes, ensuring that prefetching operations are performed efficiently. This prevents wasteful over-prefetching while maintaining the latency benefits, thereby reducing unnecessary network costs.
Solution Approach 2:
The system performs self-optimization by automatically adjusting batch sizes based on its own performance measurements. This self-service capability allows the system to minimize network costs while maintaining latency improvements without requiring external intervention or configuration.
Data Source
AI summary
A data prefetching technique optimizes the batch size of prefetch requests. The optimized batch size may be determined based on a prefetch transfer time of a previous prefetch operation, where the prefetch transfer time is measured as an elapsed time interval from when data from the previous prefetch operation is first received to when the data from the previous prefetch operation is finished being received.