Cloudwatch Custom Metrics - Combining Series


We are planning to use CloudWatch custom metrics to keep track of some customer service metrics for our phone system. We get the data from our CS provider with things like hold time, talk time, TTFR, total call time, as well as the topic etc.

We process that data and make a bunch of metrics in AWS CloudWatch (things like Total_Time with dimensions of the call type, which agent handled the call, the ring group it hit). What we’d LIKE to do is be able to make a graph that is, say, the Total_Time metric across all agents/dimensions.

The problem we have is, when we create the graph in Grafana, it always make each dimension combination a separate series. So we have hundreds of series (all possible combinations of call reasons, agents, etc…).

What’s weird is, lets say I graph out RDS database metrics in grafana. In that case, I look at CPUUtilization. If I graph the average, I get one series for each DB instance. But if I use a wildcard filter on the instance ID it gives me a single combined graph. I applied the same thinking to the custom metric, set all Dimensions to “*” but it still won’t combine the series.

Is there a way to do this that I’m missing?


See AWS CloudWatch concept: Amazon CloudWatch concepts - Amazon CloudWatch

CloudWatch does not aggregate across dimensions for your custom metrics.

So you have wrong expection/wrong metric storage, which can’t satisfy your requirements.

In theory you can use some Grafana tranformation, but I would use better metric storage.