Abstract
While raw time-series data can support various temporal analyses, such analyses are limited by the available data resolution. Storing data at the highest resolution is infeasible with data volumes. Also, computationally expensive data aggregation is necessary to serve query responses. While manual scripts can be used to adjust data resolution, this approach is tedious and not scalable. This disclosure describes techniques that use symbolic links for automatically adjusting the resolutions of time-series data in real-time based on application needs and resource availability and performance. Automatic preemptive adjustment is performed by predicting and optimizing the appropriate data resolution with machine learning models trained on historical query patterns and trends. Symbolic links to data stored at the suitable resolution(s) are provided. To serve a query response, the symbolic links are used to retrieve the data from the appropriate data store. The described techniques can be applied broadly for the storage and analysis of heterogeneous time-series data. Implementation of the techniques can optimize query handling performance, reduce computational and storage overhead, enhance scalability in adapting to changing analytical demands, and simplify overall data management for databases and applications that involve time-series data.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Mahajan, Puneet, "Automated Dynamic Adaptive Storage, Management, and Retrieval of Time-Series Data", Technical Disclosure Commons, (July 29, 2024)
https://www.tdcommons.org/dpubs_series/7242