spec.concurrency | object (nullable) | No | | Overrides of the class defaults. An absent field keeps the class value. | |
spec.concurrency.memoryShare | string (nullable) | No | | Share of the engine's memory the class may use, for example 40%. | pattern ^([1-9][0-9]?|100)%$ |
spec.concurrency.queued | integer (nullable) | No | | Queries that may wait (the resource group's queue limit). | min 0, max 10000; format int32 |
spec.concurrency.running | integer (nullable) | No | | Queries that run at the same time (the resource group's hard concurrency limit). | min 1, max 100; format int32 |
spec.concurrency.scheduling | string (nullable) | No | | How queued queries are chosen to run. | fifo, fair, weighted, weighted_fair, null |
spec.default | boolean | No | false | The class that queries land in when nothing selects another one. | |
spec.displayName | string (nullable) | No | | | max 100 |
spec.grants | array of object | No | [] | | max 50 |
spec.grants[].group | string | Yes | | | min 1, max 64; pattern ^[A-Za-z0-9][A-Za-z0-9._-]*$ |
spec.lakehouseRef | object | Yes | | The Lakehouse that hosts a catalog or a warehouse. | |
spec.lakehouseRef.name | string | Yes | | | min 1, max 63; pattern ^[a-z0-9]([-a-z0-9]*[a-z0-9])?$ |
spec.size | string | No | small | The platform's compute classes. A class is a name for four numbers (running, queued, memory share, scheduling), not a promise about hardware: the query engine has resource groups, not separate clusters. | xs, small, medium, large |