Interpreting data is an important aspect to consider when discussing improvement, along with identification. It involves not only generating data and KPIs using a tool but also understanding and evaluating those KPIs from different perspectives. Based on this evaluation, solutions can be suggested to either improve or confirm that everything is working well.

First, you have to detect and name the problems, and then the second step is finding ways to mitigate those issues. It is not about needing extra tools but more about the decision if we want to spend more manpower or not because currently, there wouldn't be any tools unless it's some kind of AI, automated solution.

Right now, the solution works quite well. I do not have any notes in terms of improvements. 

Like most Amazon products, the user interface, configuration, and tuning aren't the easiest. That's the biggest reason why people tend to go to products like TerraForm and Terragrunt. We use TerraForm and Terragrunt. So, for setting things up and interacting with X-Ray, it's definitely the user interface that can be better.

They can improve how traces are sent to other providers.

What needs to be better in AWS X-Ray is the log filtering. Predefined filters could be helpful because the power of analytics comes from how you can filter the data.

I also want to see more KPIs from AWS X-Ray.

The user interface is sometimes kind of confusing to understand. It's not very user-friendly. They should work to improve this aspect of the product. It should be easier to navigate.

While the configuration is not that difficult and you can follow the instructions in the AWS documentation, it's also not so easy to understand what exactly you have to do to configure everything. For example, in our applications, sometimes it's not clear the points I have to direct the interceptors to have the perfect trace of the endpoint. In fact, right now I'm working on a ticket here to solve a problem where the interceptor break is interrupted at a certain point due to a problem of configuration.

Between local services and AWS services, I'm working with tasks. These tasks use threads, and they are kind of getting lost due to the threads. In these cases, the configuration is a little bit difficult to understand in terms of what to do.