At Atrium Health, a nurse pulled me aside and said, "Visitors are getting angry at the kiosk." I asked why. Her answer? "The badge printer is taking 20 seconds per person. It is jammed? Buggy? No clue."
In test? It printed in 2 seconds. In staging? Same. Only production was lagging, and only with real user traffic.
Getting Serious with Dynatrace
I deployed Dynatrace OneAgent on the production kiosk. Within hours, I had visibility into:
- 📝Transaction traces of each badge print event
- 📶Network latency between kiosk, printer, and backend APIs
- 🗄️Database calls fired during check-in
Turned out, a redundant API call was re-checking photo eligibility even for returning guests, triggering slow DB joins and increasing I/O usage.
"This is the first time someone actually showed us what is happening behind that kiosk screen." — Onsite Director, Atrium Health
The Fix (and the Follow-Through)
I worked with engineering to patch the API logic: bypass unnecessary lookups for known visitors. We also cached eligibility responses at the edge. Result? 20-second waits dropped to under 3 seconds.
But I did not stop there. I helped define Dynatrace dashboards for:
- 📊Live check-in performance by hospital wing
- ⏱️Average badge print latency over time
- ⚠️Error rate breakdown by integration
Logs are great, but traces tell the full story. A patch is good. A dashboard is better.
SE Takeaways
- 🔍Logs are great, but traces tell the full story.
- 🏥Being onsite reveals friction you will never see remotely.
- 📈A patch is good. A dashboard is better.
I have never looked at a "Print Badge" button the same way since.
More from the SE Blog