Scaling LinkedIn’s search infrastructure | Rahul Raja | StaffPlus New York 2025

LinkedIn's search infrastructure evolution from a legacy system to a hosted search solution is discussed. The legacy search, utilizing SEAS and Galeen, presented operational complexities due to its custom search solutions tailored to specific applications, leading to tight coupling of platform and application logic. The hosted search solution addresses these issues by providing a managed platform where teams onboard through a Nuage service to hosted search clusters. Key components include tenant indexes, a controller for resource orchestration, and a Layer Cake machinery for runtime artifact loading, enabling multi-tenancy and clear separation of platform and application logic. This new system supports global secondary index queries in Espresso, and employs BlueGreen workflows for deployment, resulting in over 170 onboarded use cases, significantly surpassing the legacy system's onboarding rate.