Datadog

I like the unified view of synthetics for API tests and APM for code traces. I can set up a synthetic test to hit our login endpoint every minute from 5 different global locations, and if latency spikes, I can click a button to see the exact SQL query or downstream microservice that caused the delay. CI/CD integration for quality gates is also top tier, and we can block deployment if the Datadog performance baseline isn't met.

Recording browser tests is simple, reliable, and visually friendly. It has simple controls to stop and start the recording and generated test steps that were relatively easy to customize, remove, insert, and rearrange.

1st i like most about Datadog is the logging system, where it helps us to debug the data (i.e. incase the data is not integrated properly in client Point of Sale) 2nd is that we use the charts and graphs to monitor server loads and see where we can improvise and 3rd is the search functionality which we can use to find a specific data when we want to debug somethings (i.e. serach terms like trace_id, @events, etc)

The log rehydration cost and the general pricing for custom metrics is painful. We often have to sample our traces aggressively (only keeping 1% of success logs) to keep costs down, which sometimes means we miss the context on rare edge case bugs. Also, the UI for building complex dashboards can be overwhelming for new hires compared to simpler tools.

The test suite tends to have flaky failures about ~40% of the time, which makes it nearly impossible to rely on for definitive test results. Also, managing large lists of tests is very hard with the current UI and list-based system. One flow with branched logic would require one test per branch, relying on naming conventions and tagging to keep them organized, which is difficult at scale.

I dislike the most is about the cost, its too costly when the usage increases. earlier we used to log 1 month data, but since the data volume increase at our end, we had to reduce the logs to 15 days to get in budget