On-Demand Seismic Data Pre-Fetching for Storage Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional hydrocarbon reservoir modeling systems experience sub-optimal performance due to the slow retrieval of seismic data from storage mediums, leading to sluggish computation and degraded performance, especially when the dataset exceeds local memory capacity.
Innovation Solution
A system that pre-fetches defined sets of seismic data from slower storage mediums and writes them to faster local memory before performing operations, optimizing performance by minimizing repeated trace-by-trace data retrieval and maximizing storage and memory efficiency through sequential data processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If seismic data is retrieved directly from storage medium, then storage capacity is utilized, but system performance degrades due to slow retrieval speed
Solution Approach 1:
The system pre-fetches seismic data from the storage medium into local memory before it is actually needed for processing. This preliminary action of loading data ahead of time eliminates the slow retrieval bottleneck during computation, as the data is already available in fast memory when processing operations require it.
Solution Approach 2:
The patent introduces local memory as an intermediary between the storage medium and the processing unit. This intermediary layer buffers data transfers, allowing the system to utilize the high capacity of storage media while maintaining fast access speeds through the local memory cache, thus resolving the speed-capacity tradeoff.
2Productivity
If entire seismic dataset is loaded into local memory, then computation speed improves, but memory capacity is exceeded when dataset is large
Solution Approach 1:
The system divides the large seismic dataset into smaller manageable segments or chunks that fit within the available local memory capacity. It processes these segments sequentially, loading only the necessary portion into memory at any given time, thereby maintaining high computation speed without exceeding memory limits.
Solution Approach 2:
Instead of loading the entire dataset into memory, the system loads only the partial portion that is currently needed for processing. This partial action approach ensures that memory capacity is not exceeded while still achieving fast computation speeds for the active data subset.
3Quantity of substance
If data is fetched trace-by-trace from storage, then memory usage is optimized, but system performance becomes sub-optimal due to repeated retrieval
Solution Approach 1:
The system performs preliminary fetching of data segments into local memory before they are needed for processing. This pre-loading action eliminates the need for repeated trace-by-trace retrieval during computation, significantly improving system performance while maintaining efficient memory utilization through sequential processing.
Data Source
AI summary
A system to provide on-demand fetching of sequential sets of seismic data to local memory from slower storage mediums prior to performing computing operations, such as horizon auto-tracking, amplitude extraction and seismic attribute generation. As a result, the performance of the storage infrastructure (e.g., hard drives, storage filer “file server” over network, etc.) is maximized and the local memory/cache speed is fully utilized to deliver optimal system performance even when working with large seismic datasets that exceed the storage capacity of the local memory.