NAMI SERVERLESS

Our team of three software engineers collaborated remotely across North America and Europe to build Nami... As we worked through engineering our solution, we considered tradeoffs to maximize scalability and minimize complexity for the user. Here we explore some of those tradeoffs and the design decisions made during development... During load testing, we aimed for a performance of around 5,000 requests per second (RPS) to compare with real world examples that solved bursty webhook traffic issues using a serverless architecture similar to Nami. We had found that increasing the function memory size to 256MB from a default of 128MB resulted in the best response times for this scenario, optimizing scaling power. An important consideration here is that provisioning any Lambda function with higher memory has a direct correlation with longer cold start times during the container initialization process. Despite this, function execution times were still well within tolerable levels for an..