Giving your scientific computing environment (SCE) a voice
summary
Operational logs from Posit Connect, Posit Package Manager, GitLab, Kubernetes/Prometheus and AWS Batch feed a live dashboard that surfaces concrete actions: it finds 707 failed deploys and 502 unused Connect objects and proposes a Cron delete; it shows wide R version diversity and lets teams target upgrades; container telemetry reveals most interactive pods sit idle so the 16‑core default is wasteful; cost/carbon prototypes flag a user with 2,971 CPU hours in July (~$62, ~0.13 kg CO2) to guide quota and scheduling changes.
AI-written summary · about
chapters (9)
- 0:00SCE project introduction
- 1:48Operational data questions
- 6:02R versions in containers
- 7:30Optimizing idle time
- 9:38Cleaning Connect server
- 12:08Team usage patterns
- 13:58User behavior feedback
- 16:52Dashboard and user influence
- 18:25Q&A
related talks
18mData Science in Production Has Never Been So Easy
26mAccelerating Science, Decelerating Carbon: Sustainable Computing...
22mHidden Signals in K8s Clusters
33mSysAdmin or Sherlock?
27mBenchmarking and Optimizing Software Delivery Performance
20mData Science in Production: The Way to a Centralized Infrastructure
33mWhen Observability Meets Sustainability: A Real World Experience
24mAutomate your CloudOps and Spend Less Time Doing Operations