Telemetry As A Service - TaaS
"Telemetry" in SWE (Software Engineering) has been coined to record the usage metrics of a product, project. It will enable analyze the utility, efficiency, productivity, archival, performance trends over a period of time. This will provide key insights to business, analyst, engineering, infrastructure team on business utility, code/page performance, resource constraints and other overheads. The product or project as a telemetry candidate could be a Website, a mobile App, set of reports, ETLs etc. The solution provides a generic meta data driven framework to onboard and provide telemetry metrics for any application. (WEB/Apps/DB/BI/ETLs). The framework works on a subscription model where any project can subscribe to the framework. For webs or apps the subscriber would simply need to include a provided DLL to its solution. Once done, the subscriber can login onto its dashboard or use a Self Serviced PowerBI over tabular cube to access the metrics.
For a Set of reports sliced over time (year,week,month,day, hour) the framework could provide insights on :
The report concurrent hits
Peak hours
Geographical usage telemetry,
Report Side Vs Database Server side loads (rendering, processing, querying) times
Toggling statistics,
Byte/Rows Counts
Report download stats based on geography, time, format.
For a Web/Apps:
The page views
Peak hours
Geographical usage
Client process time,
send request time ,
server response (rendering, processing, querying) time
Device, Browser, operating system used
Failed Request stats
ETL :
Successful runs
Resource USage/Failure stats
Failed Stats
Average Run times
Refresh Status
This framework has onboarded couple of EDS projects on production and can be extended to provide stats/analytics for web and other source type. The fact-dimension model ensures that it is scalable and extendable for adding new features or measures. For example, for one of the projects the Data Quality checks and metrics for business rule validation dashboard has been extended with ease. Adding new measures for a product can also be done with ease.