Akamas at Performance Summit, Powered by Facebook

The Akamas team was invited to present our AI-powered performance optimization technology during the London Edition of the Performance Summit, powered by Facebook. Giovanni Gibilisco, Akamas Head of Engineering, took the virtual stage to demonstrate our novel approach to IT stack tuning: Autonomous Performance Optimization.  

Reliable, robust, and cost-effective delivery of digital services has become an essential requirement for virtually any business in any industry. On the other hand, developers, system engineering, and operations teams must contend with ever complex IT stacks.

Each technology layer comes with hundreds of user-configurable settings and cloud vendors continuously add instance type permutations to their offering. Containerization and DevOps automation and continuous integration add more complexity to the task of extracting optimal performance from application stacks.   

“Given this growing complexity, […] manual performance optimization is no longer sufficient to run these systems at their best.”

Giovanni Paolo Gibilisco, Head of Engineering @ Akamas

Many organizations resign to run applications with sub-optimal configurations, which causes degradation in performance, overprovisioning, and, ultimately, higher costs and less agility. How can you tame this complexity and make sure your systems are operating with maximum efficiency?

In this talk: 

  • We pinpoint the key challenges posed by increasing IT stack complexity and explain the status quo;
  • We show how automation can free performance engineers from tedious, time-consuming tasks; 
  • We share a customer case study where we integrated Akamas with open source monitoring and load testing tools, to improve a core online service;
  • We explain how we obtained higher throughput by 23% using 35% less memory;
  • We describe how we reduced response time degradation at peak load from 11 to 7.5 seconds.

Performance Summit was a great opportunity to share our progress with the community and compare notes with fellow performance optimization enthusiasts.