Sending Metrics to Splunk and CloudWatch: Looking for Custom Metric Integration Solutions

[rahusri2]

Hello Splunk Community,

I have a use case where we need to send metrics directly to Splunk instead of AWS CloudWatch, while still sending CPU and memory metrics to CloudWatch for auto-scaling purposes.

Datadog offers solutions, such as their AgentCheck package (https://docs.datadoghq.com/developers/custom_checks/write_agent_check/), and their repository (https://github.com/DataDog/integrations-core) provides several integrations for similar use cases. Is there an equivalent solution or approach available in Splunk for achieving this functionality?

Looking forward to your suggestions and guidance!

Thanks!

[bishida]

This sounds like a base use-case for the OpenTelemetry collector. When you run the OTel collector on an EC2, you’ll be streaming host metrics like cpu, memory, disk, and network directly to Splunk Observability Cloud.

Since the EC2 is running in AWS, you’re also able to collect most of those same metrics through CloudWatch. The big difference is the OTel collector gives you the ability to collect high-resolution streaming metrics. This is important when correlating infrastructure metrics to application performance. The metrics coming from CloudWatch will be much lower resolution by default. But yes, this approach means that metrics are going to 2 different places: Splunk Observability Cloud and CloudWatch.

The Data Management tab in Observability Cloud will give you guided instructions for installing the OTel collector.