Application Insights

AI - features

  • already a solid and widely-used solution for performance management and tracking data
  • cross-platform(.NET, NodeJS, J2EE)
  • requires only a small package to work
  • possibility to pull the data from PCs, Azure diagnostic, Docker logs etc.
  • tracks dependencies(storage, services...) out of the box
  • gathered data is available in the portal and queryable
  • integrates with VS and PowerBI

AI - usage analysis

  • available by logging custom events/traces
  • it's possible to add custom string/numeric properties so it is possible to filter the data
  • custom telemetry initializers give a possibility to e.g. perform A | B testing
  • user data is correlated thanks to a unique correlation identifier
  • it's possible to perform a constant export to a blob storage(charged additionally when used with a Basic tier). It's stored as JSON files so it is possible to perform further analysis with it

AI - caveats

  • Data is stored only for 90 days - if one needs it for a longer period, continuous export should be used
  • For a Basic tier - it's free with a 1GB limit of data per month, for an Enterprise tier it's 12,65 EUR per node for 200MB
  • For simpler scenarios is good as it is, when faced against a more complex one, it usually requires some work to code stuff
  • Throttling limits the data rate to 32 k events per second, averaged over 1 minute. The maximum cap is 500 GB/day
Made with Slides.com