Virtual Machines vs Bigstep Metal Cloud in Machine Learning
Using the H2O distributed database project, we are testing two machine learning algorithms, GLM (Generalized Linear Model) and DL (Deep Learning), in an on-premises virtualized environment and the Bigstep Metal Cloud. The benchmark analyzes the performance of both platforms for parsing the dataset and building the two models.
- Execution Times
Comparative times for parsing the data, building the GLM and DL models.
- Dataset Size Variations
Performance gaps differences when the workload size changes.
- Results Analysis
Intuition on the factors that are influencing the results