Achieve Superhuman Performance with Machine Learning
Application stacks complexity has never been so high, how machine learning can beat human experts at optimizing application performance
Life has become harder for performance engineers lately. The complexity of our stacks is increasing at an exponential pace, with an ever-growing set of middleware configurations, language runtime flags, operating system settings, and database knobs, not to mention the vast number of cloud services and deployment options made available by cloud providers.
How are we supposed to navigate this complexity and make sure we’re getting the most performance out of our stacks? Standard practices like performance modeling and testing are no longer enough, as models fail to capture complex interactions (e.g. application performance under different JVM flags or cloud instances) and performance testing can’t scale to the sheer number of combinations to evaluate.
So are we doomed? Not yet. A new powerful tool is available for performance engineers: machine learning. ML techniques are achieving unprecedented results in many application areas, however little is known about how it can be applied to optimize application performance.
We will share a new approach and solution to automatically optimize real-life applications and show how ML can explain where does application performance truly come from.
Stefano is a passionate manager based in Italy who leads the Akamas vision for performance optimization powered by machine learning. Prior to his role at Akamas, Stefano led the Research & Development business unit at Moviri, focusing on creating innovative IT performance optimization products. Before that, He delivered dozens of successful capacity and performance management consulting projects at Moviri for major national and international enterprises. He has presented several talks at the Computer Measurement Group international conference and in 2015, He won the Best Paper award for his contributions to capacity planning and performance optimization of Java applications.