@pulumi/databricks
Version:
A Pulumi package for creating and managing databricks cloud resources.
336 lines (335 loc) • 14.3 kB
TypeScript
import * as pulumi from "@pulumi/pulumi";
import * as inputs from "./types/input";
import * as outputs from "./types/output";
/**
* This resource allows you to manage [Model Serving](https://docs.databricks.com/machine-learning/model-serving/index.html) endpoints in Databricks, including custom models, external models, and foundation models. For newer foundation models, including Llama 4, please use the databricks.ModelServingProvisionedThroughput resource.
*
* > This resource can only be used with a workspace-level provider!
*
* > If you replace `servedModels` with `servedEntities` in an existing serving endpoint, the serving endpoint will briefly go into an update state (~30 seconds) and increment the config version.
*
* ## Example Usage
*
* Creating a CPU serving endpoint
*
* ```typescript
* import * as pulumi from "@pulumi/pulumi";
* import * as databricks from "@pulumi/databricks";
*
* const _this = new databricks.ModelServing("this", {
* name: "ads-serving-endpoint",
* config: {
* servedEntities: [
* {
* name: "prod_model",
* entityName: "ads-model",
* entityVersion: "2",
* workloadSize: "Small",
* scaleToZeroEnabled: true,
* },
* {
* name: "candidate_model",
* entityName: "ads-model",
* entityVersion: "4",
* workloadSize: "Small",
* scaleToZeroEnabled: false,
* },
* ],
* trafficConfig: {
* routes: [
* {
* servedModelName: "prod_model",
* trafficPercentage: 90,
* },
* {
* servedModelName: "candidate_model",
* trafficPercentage: 10,
* },
* ],
* },
* },
* });
* ```
*
* Creating a Foundation Model endpoint
*
* ```typescript
* import * as pulumi from "@pulumi/pulumi";
* import * as databricks from "@pulumi/databricks";
*
* const llama = new databricks.ModelServing("llama", {
* name: "llama_3_2_3b_instruct",
* aiGateway: {
* usageTrackingConfig: {
* enabled: true,
* },
* },
* config: {
* servedEntities: [{
* name: "meta_llama_v3_2_3b_instruct-3",
* entityName: "system.ai.llama_v3_2_3b_instruct",
* entityVersion: "2",
* scaleToZeroEnabled: true,
* maxProvisionedThroughput: 44000,
* }],
* },
* });
* ```
*
* Creating an External Model endpoint
*
* ```typescript
* import * as pulumi from "@pulumi/pulumi";
* import * as databricks from "@pulumi/databricks";
*
* const gpt4o = new databricks.ModelServing("gpt_4o", {
* name: "gpt-4o-mini",
* aiGateway: {
* usageTrackingConfig: {
* enabled: true,
* },
* rateLimits: [{
* calls: 10,
* key: "endpoint",
* renewalPeriod: "minute",
* }],
* inferenceTableConfig: {
* enabled: true,
* tableNamePrefix: "gpt-4o-mini",
* catalogName: "ml",
* schemaName: "ai_gateway",
* },
* guardrails: {
* input: {
* invalidKeywords: ["SuperSecretProject"],
* pii: {
* behavior: "BLOCK",
* },
* },
* output: {
* pii: {
* behavior: "BLOCK",
* },
* },
* },
* },
* config: {
* servedEntities: [{
* name: "gpt-4o-mini",
* externalModel: {
* name: "gpt-4o-mini",
* provider: "openai",
* task: "llm/v1/chat",
* openaiConfig: {
* openaiApiKey: "{{secrets/llm_scope/openai_api_key}}",
* },
* },
* }],
* },
* });
* ```
*
* ## Access Control
*
* * databricks.Permissions can control which groups or individual users can *Manage*, *Query* or *View* individual serving endpoints.
*
* ## Related Resources
*
* The following resources are often used in the same context:
*
* * databricks.ModelServingProvisionedThroughput to create [Foundation Model provisioned throughput](https://docs.databricks.com/aws/en/machine-learning/foundation-model-apis/deploy-prov-throughput-foundation-model-apis) endpoints in Databricks.
* * databricks.RegisteredModel to create [Models in Unity Catalog](https://docs.databricks.com/en/mlflow/models-in-uc.html) in Databricks.
* * End to end workspace management guide.
* * databricks.Directory to manage directories in [Databricks Workspace](https://docs.databricks.com/workspace/workspace-objects.html).
* * databricks.MlflowModel to create models in the [workspace model registry](https://docs.databricks.com/en/mlflow/model-registry.html) in Databricks.
* * databricks.Notebook to manage [Databricks Notebooks](https://docs.databricks.com/notebooks/index.html).
* * databricks.Notebook data to export a notebook from Databricks Workspace.
* * databricks.Repo to manage [Databricks Repos](https://docs.databricks.com/repos.html).
*
* ## Import
*
* The model serving resource can be imported using the name of the endpoint.
*
* hcl
*
* import {
*
* to = databricks_model_serving.this
*
* id = "<model-serving-endpoint-name>"
*
* }
*
* Alternatively, when using `terraform` version 1.4 or earlier, import using the `pulumi import` command:
*
* bash
*
* ```sh
* $ pulumi import databricks:index/modelServing:ModelServing this <model-serving-endpoint-name>
* ```
*/
export declare class ModelServing extends pulumi.CustomResource {
/**
* Get an existing ModelServing resource's state with the given name, ID, and optional extra
* properties used to qualify the lookup.
*
* @param name The _unique_ name of the resulting resource.
* @param id The _unique_ provider ID of the resource to lookup.
* @param state Any extra arguments used during the lookup.
* @param opts Optional settings to control the behavior of the CustomResource.
*/
static get(name: string, id: pulumi.Input<pulumi.ID>, state?: ModelServingState, opts?: pulumi.CustomResourceOptions): ModelServing;
/**
* Returns true if the given object is an instance of ModelServing. This is designed to work even
* when multiple copies of the Pulumi SDK have been loaded into the same process.
*/
static isInstance(obj: any): obj is ModelServing;
/**
* A block with AI Gateway configuration for the serving endpoint. *Note: only external model endpoints are supported as of now.*
*/
readonly aiGateway: pulumi.Output<outputs.ModelServingAiGateway | undefined>;
/**
* The Budget Policy ID set for this serving endpoint.
*/
readonly budgetPolicyId: pulumi.Output<string | undefined>;
/**
* The model serving endpoint configuration. This is optional and can be added and modified after creation. If `config` was provided in a previous apply but is not provided in the current apply, no change to the model serving endpoint will occur. To recreate the model serving endpoint without the `config` block, the model serving endpoint must be destroyed and recreated.
*/
readonly config: pulumi.Output<outputs.ModelServingConfig>;
/**
* The description of the model serving endpoint.
*/
readonly description: pulumi.Output<string | undefined>;
/**
* A block with Email notification setting.
*/
readonly emailNotifications: pulumi.Output<outputs.ModelServingEmailNotifications | undefined>;
/**
* Invocation url of the endpoint.
*/
readonly endpointUrl: pulumi.Output<string>;
/**
* The name of the model serving endpoint. This field is required and must be unique across a workspace. An endpoint name can consist of alphanumeric characters, dashes, and underscores. NOTE: Changing this name will delete the existing endpoint and create a new endpoint with the updated name.
*/
readonly name: pulumi.Output<string>;
/**
* A list of rate limit blocks to be applied to the serving endpoint. *Note: only external and foundation model endpoints are supported as of now.*
*
* @deprecated Please use AI Gateway to manage rate limits.
*/
readonly rateLimits: pulumi.Output<outputs.ModelServingRateLimit[] | undefined>;
/**
* A boolean enabling route optimization for the endpoint. *Note: only available for custom models.*
*/
readonly routeOptimized: pulumi.Output<boolean | undefined>;
/**
* Unique identifier of the serving endpoint primarily used to set permissions and refer to this instance for other operations.
*/
readonly servingEndpointId: pulumi.Output<string>;
/**
* Tags to be attached to the serving endpoint and automatically propagated to billing logs.
*/
readonly tags: pulumi.Output<outputs.ModelServingTag[] | undefined>;
/**
* Create a ModelServing resource with the given unique name, arguments, and options.
*
* @param name The _unique_ name of the resource.
* @param args The arguments to use to populate this resource's properties.
* @param opts A bag of options that control this resource's behavior.
*/
constructor(name: string, args?: ModelServingArgs, opts?: pulumi.CustomResourceOptions);
}
/**
* Input properties used for looking up and filtering ModelServing resources.
*/
export interface ModelServingState {
/**
* A block with AI Gateway configuration for the serving endpoint. *Note: only external model endpoints are supported as of now.*
*/
aiGateway?: pulumi.Input<inputs.ModelServingAiGateway>;
/**
* The Budget Policy ID set for this serving endpoint.
*/
budgetPolicyId?: pulumi.Input<string>;
/**
* The model serving endpoint configuration. This is optional and can be added and modified after creation. If `config` was provided in a previous apply but is not provided in the current apply, no change to the model serving endpoint will occur. To recreate the model serving endpoint without the `config` block, the model serving endpoint must be destroyed and recreated.
*/
config?: pulumi.Input<inputs.ModelServingConfig>;
/**
* The description of the model serving endpoint.
*/
description?: pulumi.Input<string>;
/**
* A block with Email notification setting.
*/
emailNotifications?: pulumi.Input<inputs.ModelServingEmailNotifications>;
/**
* Invocation url of the endpoint.
*/
endpointUrl?: pulumi.Input<string>;
/**
* The name of the model serving endpoint. This field is required and must be unique across a workspace. An endpoint name can consist of alphanumeric characters, dashes, and underscores. NOTE: Changing this name will delete the existing endpoint and create a new endpoint with the updated name.
*/
name?: pulumi.Input<string>;
/**
* A list of rate limit blocks to be applied to the serving endpoint. *Note: only external and foundation model endpoints are supported as of now.*
*
* @deprecated Please use AI Gateway to manage rate limits.
*/
rateLimits?: pulumi.Input<pulumi.Input<inputs.ModelServingRateLimit>[]>;
/**
* A boolean enabling route optimization for the endpoint. *Note: only available for custom models.*
*/
routeOptimized?: pulumi.Input<boolean>;
/**
* Unique identifier of the serving endpoint primarily used to set permissions and refer to this instance for other operations.
*/
servingEndpointId?: pulumi.Input<string>;
/**
* Tags to be attached to the serving endpoint and automatically propagated to billing logs.
*/
tags?: pulumi.Input<pulumi.Input<inputs.ModelServingTag>[]>;
}
/**
* The set of arguments for constructing a ModelServing resource.
*/
export interface ModelServingArgs {
/**
* A block with AI Gateway configuration for the serving endpoint. *Note: only external model endpoints are supported as of now.*
*/
aiGateway?: pulumi.Input<inputs.ModelServingAiGateway>;
/**
* The Budget Policy ID set for this serving endpoint.
*/
budgetPolicyId?: pulumi.Input<string>;
/**
* The model serving endpoint configuration. This is optional and can be added and modified after creation. If `config` was provided in a previous apply but is not provided in the current apply, no change to the model serving endpoint will occur. To recreate the model serving endpoint without the `config` block, the model serving endpoint must be destroyed and recreated.
*/
config?: pulumi.Input<inputs.ModelServingConfig>;
/**
* The description of the model serving endpoint.
*/
description?: pulumi.Input<string>;
/**
* A block with Email notification setting.
*/
emailNotifications?: pulumi.Input<inputs.ModelServingEmailNotifications>;
/**
* The name of the model serving endpoint. This field is required and must be unique across a workspace. An endpoint name can consist of alphanumeric characters, dashes, and underscores. NOTE: Changing this name will delete the existing endpoint and create a new endpoint with the updated name.
*/
name?: pulumi.Input<string>;
/**
* A list of rate limit blocks to be applied to the serving endpoint. *Note: only external and foundation model endpoints are supported as of now.*
*
* @deprecated Please use AI Gateway to manage rate limits.
*/
rateLimits?: pulumi.Input<pulumi.Input<inputs.ModelServingRateLimit>[]>;
/**
* A boolean enabling route optimization for the endpoint. *Note: only available for custom models.*
*/
routeOptimized?: pulumi.Input<boolean>;
/**
* Tags to be attached to the serving endpoint and automatically propagated to billing logs.
*/
tags?: pulumi.Input<pulumi.Input<inputs.ModelServingTag>[]>;
}