How We Scale
Duckling Sizes
MotherDuck’s per-user tenancy model gives each user an isolated (DuckDB instance) in one of five sizes to enable individual, user-level configuration.
Pulse
Our smallest instance, perfect for ad-hoc analytics tasks
Standard
Built to handle common data warehouse workloads, including loads and transforms
Jumbo
For larger data warehouse workloads with many transformations or complex aggregations
Mega
An extremely large instance for when you need complex transformations done quickly
Per-user tenancy and vertical scaling
MotherDuck employs a per-user tenancy and vertical scaling strategy. Users connect to their own MotherDuck Ducklings (DuckDB instances), which are sized (pulse, standard, jumbo, mega, giga) to meet their specific needs. There is also the option for additional Ducklings, through read scaling (explained below), to ensure flexible resource allocation. Ultimately, each Duckling establishes a connection with the central Data Warehouse storage.
Read Scaling
MotherDuck's read scaling capabilities allow users to connect via a BI Tool to dedicated Ducklings that function as read replicas. These read replicas can be provisioned in various sizes (pulse, standard, jumbo, mega or giga) to accommodate different needs. Ultimately, these read replicas connect to the Data Warehouse storage, enabling efficient handling of read operations.
