All classes are under active development and subject to non-backward compatible changes or removal in any future version. These are not subject to the Semantic Versioning model. This means that while you may use them, you may need to update your source code when upgrading to a newer version of this package.
| Language | Package |
|---|---|
 TypeScript |
@cdklabs/generative-ai-cdk-constructs |
 Python |
cdklabs.generative_ai_cdk_constructs |
Thanks to @jimini55, @scoropeza, @PaulVincent707, @Ishanrpatel, @lowelljehu and @rddefauw for the initial version of this construct.
This construct provides an Amazon CloudWatch dashboard to monitor metrics on Amazon Bedrock models usage. The specific list of metrics created by this construct is available here.
Note: Native metrics for Amazon Bedrock don't support dimensions beyond model ID. If a single account is hosting multiple workloads in the same region, the Bedrock metrics would be aggregated across all workloads.
Here is a minimal deployable pattern definition:
TypeScript
import { Construct } from 'constructs';
import { Stack, StackProps, Aws } from 'aws-cdk-lib';
import { BedrockCwDashboard } from '@cdklabs/generative-ai-cdk-constructs';
const bddashboard = new BedrockCwDashboard(this, 'BedrockDashboardConstruct', {});
// provides monitoring for a specific model
bddashboard.addModelMonitoring('claude3haiku', 'anthropic.claude-3-haiku-20240307-v1:0', {});
// provides monitoring of all models
bddashboard.addAllModelsMonitoring({});Optionally, you can also use the Bedrock models to access the modelId:
import { bedrock, BedrockCwDashboard } from '@cdklabs/generative-ai-cdk-constructs';
...
// provides monitoring for a specific model
bddashboard.addModelMonitoring(
'claude3haiku',
bedrock.BedrockFoundationModel.ANTHROPIC_CLAUDE_HAIKU_V1_0.modelId,
{}
);
...Python
from constructs import Construct
from cdklabs.generative_ai_cdk_constructs import BedrockCwDashboard
bddashboard = BedrockCwDashboard(self, 'BedrockDashboardConstruct')
# provides monitoring for a specific model
bddashboard.add_model_monitoring(
model_name: 'claude3haiku',
model_id: 'anthropic.claude-3-haiku-20240307-v1:0'
)
# provides monitoring of all models
bddashboard.add_all_models_monitoring()new BedrockCwDashboard(scope: Construct, id: string, props: BedrockCwDashboardProps)
Parameters
- scope Construct
- id string
- props BedrockCwDashboardProps
| Name | Type | Required | Description |
|---|---|---|---|
| existingDashboard | aws_cloudwatch.Dashboard |  |
Existing dashboard to be used by the construct. Mutually exclusive with dashboardName - only one should be specified. |
| dashboardName | string | |
A name for the dashboard which will be created. If not provided, the construct will create a new dashboard named 'BedrockMetricsDashboard'. Mutually exclusive with existingDashboard - only one should be specified. |
| Name | Type | Description |
|---|---|---|
| dashboard | aws_cloudwatch.Dashboard | The CloudWatch Dashboard used by the construct (whether created by the construct or provided by the client) |
Provide metrics for a specific model id in Bedrock
@param {string} modelName - Model name as it will appear in the dashboard row widget.
@param {string} modelId - Bedrock model id as defined in https://docs.aws.amazon.com/bedrock/latest/userguide/model-ids.html
@param {ModelMonitoringProps} props - user provided props for the monitoring.
Add a new row to the dashboard providing metrics across all model ids in Bedrock
@param {ModelMonitoringProps} props - user provided props for the monitoring.
Out-of-the-box implementation of the construct without any override will set the following defaults:
- Dashboard name is
BedrockMetricsDashboard
- Period (the period over which the specified statistic is applied) is set to one hour
- The following metrics are displayed for the model specified:
- InputTokenCount
- OutputTokenCount
- InvocationLatency (min, max, average)
- Invocations (sample count)
- InvocationClientErrors
- Period (the period over which the specified statistic is applied) is set to one hour
- The following metrics are displayed for all models:
- InputTokenCount
- OutputTokenCount
- InvocationLatency (min, max, average)
- Invocations (sample count)
- InvocationClientErrors
You are responsible for the cost of the AWS services used while running this construct.
We recommend creating a budget through AWS Cost Explorer to help manage costs. Prices are subject to change. For full details, refer to the pricing webpage for each AWS service used in this solution:
When you build systems on AWS infrastructure, security responsibilities are shared between you and AWS. This shared responsibility model reduces your operational burden because AWS operates, manages, and controls the components including the host operating system, virtualization layer, and physical security of the facilities in which the services operate. For more information about AWS security, visit AWS Cloud Security.
Optionnaly, you can provide existing resources to the constructs (marked optional in the construct pattern props). If you chose to do so, please refer to the official documentation on best practices to secure each service:
If you grant access to a user to your account where this construct is deployed, this user may access information stored by the construct (Amazon CloudWatch logs). To help secure your AWS resources, please follow the best practices for AWS Identity and Access Management (IAM).
AWS CloudTrail provides a number of security features to consider as you develop and implement your own security policies. Please follow the related best practices through the official documentation.
This solution depends uses the Amazon Bedrock and Amazon CloudWatch services, which are not currently available in all AWS Regions. You must launch this construct in an AWS Region where these services are available. For the most current availability of AWS services by Region, see the AWS Regional Services List.
Note You need to explicity enable access to models before they are available for use in Amazon Bedrock. Please follow the Amazon Bedrock User Guide for steps related to enabling model access.
Service quotas, also referred to as limits, are the maximum number of service resources or operations for your AWS account.
Make sure you have sufficient quota for each of the services implemented in this solution. For more information, refer to AWS service quotas.
To view the service quotas for all AWS services in the documentation without switching pages, view the information in the Service endpoints and quotas page in the PDF instead.
© Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.