Datadog Integration Using Cloudwatch Metrics Stream

Background

AWS developed a streaming solution to forward Cloudwatch metrics to third-party providers. It promises to reduce the Datadog metrics delay significantly.

Who is this announcement for?

For all teams that are using Datadog to monitor AWS resources metrics, e. g. RDS, ElastiCache, etc.

Why do I need this?

For most AWS resources, except for EC2 and ECS metrics that sent through an Datadog agent, there is a polling delay of 10 minutes before the metrics is available in Datadog. Combined with 2-5 mins delay of metrics available in Cloudwatch, total delay is around 12-15 mins.

Stream metrics to Datadog can reduce the delay to 3-4 mins with consistent 1 minute periodic update.

What are the trade-off?

AWS cost of sending metrics to Datadog will increase
The increase is vary between services, you can check the documentation linked below for the detail.

How do we start?

More detail on Cloudwatch Metrics Stream can be found in this page
Guide to implement the Terraform module can be found in this page
Contact @igit or @stevensteven for any follow up questions

Thank you!