API Analytics

[chriswininger]

We need more visibility into the health and usage of the API, from both an operations standpoint and a business standpoint:

Things we would like to track:

• Requests per/minute/second/day etc... for each end-point

• Identify information (depends on implementing identity keys); this will likely come later

Considerations:

• This should be easy to implement and maintain
• It should be easy for us to access and view the data in a visual form
• Access should not be open to the public

Some Ideas/ frameworks to look at:

• graphana? DataDog? Paper Trail? Built in Azure capabilities? Node frameworks? https://github.com/electerious/Ackee? Jupyter notebooks

[chriswininger]

Currently Investigating a graphana solution:

https://medium.com/@chamerling/getting-metrics-from-your-express-app-without-effort-with-docker-and-grafana-ac8f6c42cbfb

This is more of a role your own approach but does a fair amount for us

There is also Azure Application Insights (AI):

https://youtu.be/jCfJWYocz1M https://youtu.be/pqZF8LjBh68

This would be available if we switched to hosting our application as an Azure Web Application. This sounds a bit like AWS Fargate and may be something we should consider:

Nginx Layer:

https://medium.com/faun/total-nginx-monitoring-with-application-performance-and-a-bit-more-using-8fc6d731051b

We could also try to extract the metrics from nginx which may actually capture some things we'll miss at the node level, but this would go away if we switched to Azure Web Application