The API reference manual introduces the data model of Batch Compute and interfaces (APIs) used for cluster, job, and image management. Before you use Batch Compute APIs, make sure that you understand the product description, user agreements, billing methods, SDKs, and quick start of Batch Compute. For more information, see What is Batch Compute.
Batch Compute provides you with the cluster, job, and image management functions, including:
Creates, deletes, queries, and lists clusters.
Submits, stops, restarts, queries, and lists jobs.
Queries and lists tasks.
Queries and lists instances.
Terms
Term | Description | Naming rule |
---|---|---|
Cluster | Virtual cluster, or ECS instant group. Management unit of ECS instances. | It can contain letters, numbers, hyphens (-), and underscores (_). Do not start with numbers. The length can be [1, 64] bytes. |
Job | A specific computing need of a user. | It can contain letters, numbers, hyphens (-), and underscores (_). Do not start with numbers. The length can be [1, 64] bytes. |
Task | Composed of a group of tasks and dependencies. Jobs can be described in the form of directed acyclic graphs (DAGs). | It can contain letters, numbers, hyphens (-), and underscores (_). Do not start with numbers. The length can be [1, 32] bytes. |
Instance | Each task may contain one or more instances. Instance is the minimum unit in the scheduling and execution of Batch Compute. Instances run on the nodes of the Batch Compute distributed system. | N/A |
Limits
In Batch Compute, the number of virtual clusters you can create is limited.
In the API description section, whenever a conflict arises between optional parameter values or available specifications and specification constraints given on the official website, the information on the official website prevails.