DB Testing
TRACK TALK
Replaying Logs as a Load Profile for MongoDB: Myth or Reality?
This talk explores the practical journey of using application MongoDB logs to conduct capacity testing. Faced with the limitations of synthetic load generation, her team embarked on an ambitious project to develop a tool that could accurately replay production traffic against MongoDB. She will dive into the methodology, the significant challenges encountered, the solutions developed, and the ultimate conclusions on whether log replay is a viable and effective approach for performance and capacity testing. This talk will share crucial insights, lessons she and her team learned.
In this talk, Raisa will share the journey of building a MongoDB log replay framework for performance and capacity testing. She will introduce the motivation behind the project, discuss why synthetic load generation was insufficient, and explain how her team chose log replay as a solution. The talk will cover the proof of concept, anonymisation of sensitive data, log distribution, handling complex operations, preserving traffic patterns, and scaling scenarios including failover and warm-up. Raisa will also present key metrics, evolved architecture, surprising discoveries, and lessons learned, concluding with insights into when log replay is most effective.
What you’ll learn
Session details

Raisa Lipatova
Raisa has over 15 years of experience in high-tech, working across QA engineering, automation, performance testing, and team leadership. Raisa career has taken her from hands-on testing to leading Performance team and cross-functional projects. Currently, She’s focused on developing more technical system architecture things and deep dive into performance engineering. Also She’s creating educational resources and communities for SW engineer professionals, sharing her knowledge to help others grow