Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

First, centralized logging is not just a good idea -- its key when you start working with multiple servers (which will most likely be almost right away). You need to be able to trace requests / responses / errors across your platform. Many tools (including logging library -> database and a custom log search / viewer) can give you this. Just pick something that works for your budget and development process and start there. To track a single chain of events, you'll just need to have a GUID that you pass between calls in a single request (used for logging).

Next, you'll want to track analytics centrally. Etsy and Netflix have been pioneers in this area. Their engineering blogs are very good to follow. Think: something like a timeseries database (like Influx / Prometheus) and getting data into it. Use tools like Grafana to get data out of it in dashboards or reports. This is separate from your application debug / error logging system.

The next step after this is developing something that consumes data from both of those systems and provides alerts based on unusual activity -- something that provides early warning to devops.



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: