At Pivotus we are building a secure messaging application for financial institutions. Last fall our team made a commitment to the Reactive approach using Lagom servicing React and React Native clients. This talk will focus on performance testing and metrics for exercising Reactive at scale. We will review the performance infrastructure, demo a sample scenario, and cover reporting and visualization examples during this talk. Tools used include Jenkins, Jmeter, Grafana and InfluxDB.
For performance testing we started by composing primitive tests to find high watermarks per API for each 9 Reactive services, running on a single pod. After establishing the baselines, we ran these tests on the realistic multi-pod deployment, then examined realistic use cases with multiple API calls. We set up endurance testing to analyze the system behavior during the long-term load, they run independently or in conjunction with manual testing/automation.
SHARE THIS TALK
Pivotus Ventures, Director of Engineering
Lilit Yenokyan has BS in Applied Math from Yerevan State University and MS in Computer Science from University of Michigan She started her career as software engineer and worked at number of startups. Her main focus has later shifted to architecting automation frameworks. Lilit later worked at Online Services department at Microsoft developing editorial tools.
In 2013 Lilit joined Netflix as Senior Software Engineer. There she worked on number of teams, focusing on improving video delivery and bringing to Netflix users key features such as mobile downloads. Netflix nurtured Lilit's curiosity of chasing inventions and doing things no one imagined possible before. Last fall, Lilit joined Silicon Valley startup, Pivotus Ventures, as Director of Engineering, where she leads teams responsible for software delivery pipeline, testing and infrastructure. Testing Reactive microservices performance at scale is one of Lilit's core focus areas. She is a frequent speaker at technical conferences and events.