Why Metrics are the Most Critical Data Type in Observability

Editor’s Note: This is the fourth and final installment of a series of blog posts previewing our State of Observability 2024 survey report.

In last week’s episode of this blog series, we looked at whether observability is replacing or enhancing existing IT monitoring tools. This week, we’ll look at why metrics are the most important observability data type to ITOps teams and what's holding back tracing.

Metrics Are the Most Critical Observability Data Type

Observability data typically refers to data generated from metrics, events, logs and traces. So, we asked our survey respondents which of these data types was most critical in their observability strategies. The winner by a wide margin was metrics, cited by 64% of respondents. Metrics in observability typically includes traditional KPIs such as

CPU and memory usage
System error rate
Application or service request rate

Some of the more advanced metrics that are becoming increasingly popular are:

Cloud cost calculations
Power consumption
Sustainability measurements.

It’s this combination of traditional system metrics that IT organizations rely on every day and more advanced metrics that help IT solve its more modern problems that make metrics the most critical data type in observability.

Logs and events came in around the middle of the pack, at 56% and 55% respectively. Traces, the newest observability data type trailed the rest of the pack at just 33%.

Blog_Report_Observability__2024_P4_Chart_01-100

Just 27% of enterprises told us traces were the most critical data type.

Blog_Report_Observability__2024_P4_Chart_02_01-100-1

Why are Traces Lagging in the Survey Responses?

As you can see from the charts above, traces are a much less critical data type in observability for both enterprises and MSPs. Our next question answered why. We asked our survey respondents what percentage of their business-critical applications are built to generate “trace” data. These are typically serverless or microservices-based applications. Most respondents—60%--said 30% of less of their business-critical applications generated trace data. Less than 10% of respondents said more than 50% of their business-critical applications generated trace data.

While microservices and serverless may be the wave of the future, organizations we surveyed are still adopting these application architectures gradually and for now, have a lesser need for distributed tracing vs. other observability data types.

Blog_Report_Observability__2024_P4_Chart_03-100

With that, we conclude this blog series. There are many more data points and insights in the full State of Observability 2024 report for both enterprises and MSPs to consider as they assess their observability strategies.