Adaptive Data Placement Framework Optimizes Data Storage Costs in Multicloud Environments

Strategic Insights -

Researchers at the Copenhagen Business School have proposed an adaptive data placement framework (ADAPT) that optimizes data storage costs and enhances data availability in multicloud environments. The framework, which has been peer-reviewed, integrates machine learning models to select optimal storage locations and improve data file availability. In a study published in Cluster Computing, the researchers demonstrated the effectiveness of the ADAPT framework in reducing data storage costs and increasing data availability.

Key Takeaways:

  • The ADAPT framework is designed to optimize data storage costs and enhance data availability in multicloud environments.
  • The framework uses machine learning models to select optimal storage locations and improve data file availability.
  • The XGBoost model effectively improves cost efficiency from 6.58% to 24.26% and data file availability.
  • The ADAPT framework has been integrated with four popular machine learning models, including XGBoost, to evaluate its performance.
  • The study demonstrated the effectiveness of the ADAPT framework in reducing data storage costs and increasing data availability in multicloud environments.

Statistics:

  • The ADAPT framework improves cost efficiency by 17.68% (24.26% - 6.58%).
  • The XGBoost model reduces data storage costs by 6.58% to 24.26%.
  • The study applied four popular machine learning models, including XGBoost, to evaluate the performance of the ADAPT framework.
  • The ADAPT framework has been peer-reviewed and published in Cluster Computing, a journal published by Springer.

Sources:

  • "Adapt: an Effective Data-aware Multicloud Data Placement Framework." Cluster Computing, 2025;28(13).
  • Springer (www.springer.com; Cluster Computing - www.springerlink.com/content/1386-7857/)
  • Somnath Mazumdar, Copenhagen Business School, Dept Digitalisat, Solbjerg Plads 3, DK-2000 Frederiksberg, Denmark.
  • John Agyekum and Christoph Scheich, additional authors for this research.