One of the datapoints our organization is asking us for is, “give me the throughput by pod?” Most people think, let me go to the service view because it will tell me the throughput (or failure rate/response time) by pod but that isn’t the case. Then they go to the K8s app where that information doesn’t exist either. Then they come to our internal Dynatrace team, and we have to show them how to query for it. It’s not exactly an easy data point to find without having to a) query for it or b) split by in the distributed traces app. Even both of those options pose problems because you are targeting trace data, which has low retention rate so to solve that, you’d have to create a pipeline and turn that data into metrics which isn’t a great answer either.
This feels like it should be easy, out-of-the-box data in the new service view. When I look at the new service view, there is an “infrastructure” tab that gives me a table of CPU and Mem usage by pod for that service, this is exactly what I want, but for throughput, response time, and failure rate. Taking it one step further.. a toggle between table and time chart views for all of this data.
There are many different mesh/load balancing products it’s important to understand if load is evenly distributed within our application. Load balancing also affects how we set our HPAs. Some teams even scale based on throughput. Throughput by pod isn’t the only ask, it would be ideal to also quickly visualize response time and failure rate by pod in the default service view, without it being multiple clicks or queries away. Not having this information readily available in the service view is driving users to create custom dashboards rather than paying attention to the default Dynatrace views.
