Right now we mostly use it for log-to-trace navigation and basic service health metrics (like error rates, p95 latency, and overall request volume).
jasondjk 20 hours ago [-]
I’d contend having user behaviour logged directly into your monolithic app db using the ahoy gem instead of some external grafana/etc system is much more powerful because then you can point an agent at a copy of your database and it can wire together all of your operational and business metrics at the same time without having to try and say, connect a user profile between grafana/salesforce/hubspot/intercom etc. eg. “Claude, tell me which customers activate and convert and which churn and why and make a plan to optimise the funnel”. Hooking that up has never been easier https://insidertrades.directory/built-with-rails/ahoy-blazer... ie. build, don’t buy in the age of AI
drdexebtjl 19 hours ago [-]
No, the right way is to just push the transactional data into your analytical database periodically, and then point the model at the analytical database.
On every place I worked, logs, traces and metrics represent approximately 100% of their data by size. The transactional data is a rounding error.
__vivek 12 hours ago [-]
Ahoy tracks user behavior, but it can't give you spans. If you want to know exactly how many milliseconds an external API or DB query took, you need OTel.
see: https://grafana.com/docs/loki/latest/send-data/otel/
On every place I worked, logs, traces and metrics represent approximately 100% of their data by size. The transactional data is a rounding error.