Integrate Amazon CloudWatch Metric Streams
Dynatrace integration with Amazon CloudWatch Metric Streams provides a simple and safe way to ingest AWS metrics. Amazon CloudWatch Metric Streams allows all metrics issued in a given AWS region to be streamed through Kinesis Firehose to the Dynatrace API.
AWS default integration vs AWS Metric Streaming
Differences between AWS default integration and AWS Metric Streaming.
AWS Default integration | AWS Metric Streaming | |
---|---|---|
ActiveGate | Required for non-built-in services or monitored environments bigger than 2000 instances️ | Not required️ |
Kinesis Firehose | Not required️ | Required️ |
Public endpoint | Not required️ | Required |
Available metrics | Selected Amazon CloudWatch metrics️ | All available Amazon CloudWatch metrics️ |
Metrics selection | In Dynatrace️ | In Amazon CloudWatch console️ |
Metrics selection scope | Monitored metrics selection possible at the level of a single metric and its statistics️ | Monitored metrics selection possible only at the level of the whole namespace️ |
Metrics key prefix | ext:cloud.aws.<service> ️ | cloud.aws.<service> ️ |
Dynatrace Entities (topology attributes and tags) | ️ | |
Predefined alerts | ||
Predefined dashboards |
Prerequisites
- Create an API token in your Dynatrace environment and enable the Ingest metrics permission.
- Determine the API URL for your environment:
- For Dynatrace SaaS
https://<your_environment_ID>.live.dynatrace.com
- For Dynatrace Managed
https://<your_domain>/e/<your_environment_ID>
- For ActiveGate
https://<your_active_gate_IP_or_hostname>:9999/e/<your_environment_ID>
- For Dynatrace SaaS
To determine <your_environment_ID>
, see environment ID.
You need to have your Dynatrace tenant open to incoming Internet traffic (or at least to AWS traffic) to be able to receive the AWS metrics. Restrictive firewalls might block the streaming service.
Set up a Metric Streams client
You can set up a Metric Streams client with a CloudFormation template or in the AWS console. See below for instructions.
CloudFormation allows you to deploy a Metric Streams client using a single deployment command to create a stack that groups multiple AWS resources. This approach is faster and makes AWS resource management easier.
You need one client stack for each region you want to monitor. Once deployed, the client starts streaming all metrics produced in its region. You can restrict which metrics are streamed.
Deploy the Metric Streams client for a default region
To fetch the CloudFormation template and deploy it to your AWS account, run the command below. Be sure to replace <your_API_URL>
and <your_API_token>
with your own values. Consult the parameters table that follows for details.
If you have AWS CLI configured, you can use a Bash-compliant shell. Otherwise, you can use CloudShell, which is available in the AWS console.
DYNATRACE_ENV_URL=<your_API_URL>
DYNATRACE_API_KEY=<your_API_token>
STACK_NAME=dynatrace-aws-metric-streams-client
DELIVERY_ENDPOINT=https://aws.cloud.dynatrace.com/
REQUIRE_VALID_CERTIFICATE=true
wget -O dynatrace-aws-metric-streams-client.yaml https://assets.cloud.dynatrace.com/awsmetricstreaming/dynatrace-aws-metric-streams-client.yaml && \
aws cloudformation deploy --capabilities CAPABILITY_NAMED_IAM --template-file ./dynatrace-aws-metric-streams-client.yaml --stack-name $STACK_NAME --parameter-overrides DynatraceEnvironmentUrl=$DYNATRACE_ENV_URL DynatraceApiKey=$DYNATRACE_API_KEY RequireValidCertificate=$REQUIRE_VALID_CERTIFICATE FirehoseHttpDeliveryEndpoint=$DELIVERY_ENDPOINT
Deploy the Metric Streams client for other regions
The command above uses the default AWS CLI profile and its default region. To change the profile and region, you can export additional variables such as AWS_DEFAULT_REGION
and AWS_PROFILE
and rerun the deployment command. If you are using CloudShell, you can change the region in the AWS console instead. For details on how to configure the AWS CLI, see Environment variables to configure the AWS CLI.
Confirm that the Metric Streams client was deployed correctly optional
To ensure that the Metric Streams client was deployed correctly
- In the AWS console, go to CloudFormation.
- Select the stack you created in the CloudFormation deployment.
- On the Events tab, make sure all events completed successfully and there are no failed events.
- In Parameters tab, make sure all parameters you provided have correct values.
Restrict which metrics are streamed
If you want to restrict which metrics are streamed
- In the AWS console, go to CloudFormation.
- Select the stack you created in the CloudFormation deployment.
- On the Resources tab, find the resource with type
AWS::CloudWatch::MetricStream
and note its Physical ID. - Go to CloudWatch.
- Under Metrics, select Streams.
- In the list of metric streams, select the one whose Name corresponds to the Physical ID that you noted in step 3 and then select Edit.
- Under Metrics to be streamed, select one of the following options:
- All namespaces, if you want to automatically stream all namespaces (you can manually select namespaces to exclude).
- Selected namespaces, if you want to manually select the namespaces to stream.
- Under Select metrics for the metric stream, select one of the following options:
- All metrics, if you want to automatically stream all metrics from namespaces selected in step 7.
- Exclude metrics by metric name, if you want to manually exclude metrics for each namespace.
- Select Save changes.
If you don't have access to the CloudFormation template, you can manually set up a Metric Streams client in the AWS console. Follow the steps below for instructions.
Create a Kinesis Data Firehose stream
Create a CloudWatch Metric Stream
For each region that you want to monitor, you need to repeat the entire procedure: create a Kinesis Data Firehose stream and create a CloudWatch Metric Stream.
Create a Kinesis Data Firehose stream
- In the AWS console, go to Kinesis.
- Select Create delivery stream.
.
Name and source
- Enter a stream name and save it for later use.
- In Source, make sure Direct PUT or other sources is selected.
- Select Next.
.
Process records
- Make sure Data Transformation and Record format conversion are disabled.
- Select Next.
.
Choose a destination
-
Select Third-party service provider and choose Dynatrace.
-
In HTTP endpoint URL, select one of available Dynatrace endpoints (Global, EU, US).
-
In Access key, enter your API token. See Prerequisites for instructions.
-
In Content encoding, make sure GZIP is selected.
-
optional In Parameters, select Add parameter and enter the parameters and their values as follows:
Parameter Description dt-url
Your API URL. See Prerequisites for instructions. require-valid-certificate
Enter true
orfalse
. If set totrue
, Dynatrace verifies the SSL certificate of your Dynatrace environment URL. -
In Retry duration, enter
900
. -
In S3 backup mode, make sure Failed data only is selected.
-
In S3 bucket, select Create new.
-
Enter a name and (optionally) a region for the S3 bucket, and then select Create S3 bucket.
-
Select Next.
.
Configure settings
- Use the existing default settings.
- optional In Tags, enter tags to organize your AWS resources.
- Select Next.
.
Review
- Review your configuration.
- Select Create delivery stream.
Create a CloudWatch Metric Stream
- In the AWS console, go to CloudWatch.
- Under Metrics, select Streams.
- Select Create metric stream.
- Select one of the options below:
- All namespaces, if you want to automatically stream all namespaces (you can manually select namespaces to exclude).
- Selected namespaces, if you want to manually select the namespaces to stream.
- Under Select metrics for the metric stream, select one of the following options:
- All metrics, if you want to automatically stream all metrics from namespaces selected in step 7.
- Exclude metrics by metric name, if you want to manually exclude metrics for each namespace.
- In Configuration, select Select an existing Firehose owned by your account and enter the name of the Firehose stream that you created in the previous section.
- In Change output format, make sure Open Telemetry 0.7 is selected.
- In Metric stream name, enter a name for your metric stream.
- Select Create metric stream.
View metrics using preset dashboards
Once you deploy the Metric Streams client, you can use the predefined dashboards from the GitHub repository in the Dynatrace web UI to visualize your ingested data.
To upload preset dashboards from GitHub
- Get
upload_dashboards.py
from the GitHub repository.
curl -o upload_dashboards.py https://raw.githubusercontent.com/Dynatrace/snippets/master/product/dashboarding/upload_dashboards.py
-
Create a
dashboards
directory next toupload_dashboards.py
. -
Add any dashboard definition from GitHub to your
dashboards
directory.Each dashboard definition is a single JSON file located in the folders of the GitHub repository.
-
Run the script below. Be sure to replace
<your_dynatrace_cluster_version>
,<your_API_token>
, and<your_API_URL>
with your own values. Consult the parameters table that follows for details.
python3 upload_dashboards.py --cluster-version <your_dynatrace_cluster_version> --dynatrace-api-token <your_API_token> --dynatrace-env-url <your_API_URL>
Example command
python3 upload_dashboards.py --cluster-version 206 --dynatrace-api-token 123456789 --dynatrace-env-url https://my-cluster.com/e/1755ddb2-7938-41a2-b6bd-096e0fdcd3e0
Uninstall the Metric Streams client
If you deployed the Metric Streams client with a CloudFormation template
- In the AWS console, go to CloudFormation.
- Select the stack you created in the CloudFormation deployment.
- In Resources, find the resource with type
AWS::S3::Bucket
, select its link, and, in the S3 console, delete all objects in this bucket. - Back in CloudFormation, in Stack information, select Delete.
If you deployed the Metric Streams client through the AWS console, delete all the resources you created (S3 bucket, Kinesis Firehose delivery stream, CloudWatch metric stream).
Metric Streams limitations
A metric won't be streamed if it is more than two hours old. You can determine a metric's age by graphing it in the CloudWatch console and checking the age of the last datapoint displayed.
For more limitations, see Amazon CloudWatch troubleshooting page