Multi-tenant applications face the unique challenge of scaling along two dimensions: the number of tenants and the workload of each tenant. These applications must scale while maintaining consistent low latency and efficient resource utilization. Existing solutions range from giving each tenant its own database to consolidating all tenants within a single database, schema, and tables. The former offers flexibility and isolation at the expense of efficiency and engineering complexity, while the latter is efficient and simple but risks noisy-neighbor problems and data leaks.
In this talk, I’ll share how Nile approached adapting Postgres for multi-tenant applications. We’ll share the overall architecture and dive into key decisions. We’ll share how we leveraged existing Postgres capabilities and introduced a new purpose-built storage layer that allows us to isolate tenants and migrate them between Postgres instances in order to scale to millions of tenants.
Pizza is provided, and we also have a user group lending library of books & papers related to PostgreSQL and Databases.
NEW LOCATION: same place as the summer BBQ - the Fred Hutch Steam Plant building. (Entrance on Eastlake Ave) Typically it's not hard to find street parking nearby.
