Engineering at Scale: The Challenges of Predicting Queries in Web Search Engines


Paul Baecke


The topic will be building a high availability, low-latency global service that powers autocompletion across multiple end-points. This will cover a general overview of the architecture and some reasons why specific choices were made in the architecture. There will also be example of code snippets that lead to ‘interesting behaviour’ when scaled up to 400,000+ requests per second, where stand-alone code optimizations behave badly at scale.