What metrics can we extract from our container based platforms (e.g. K8s logs) and Posit products (e.g. Connect) to better optimise our platforms for cost, compliance and user experience?

[epijim]

Proposal

Over the last few months we've realised a few questions can be answered from our SCEs operational data (e.g. promotheus logs, Connect and Github APIs, stc) like:

• How is our user base split across official and un-officially supported departments?
• How much content on our Connect server never successfully deployed, or hasn't been looked at in X years?
• When given the option of different R versions (as containers), how quickly do our users migrate to new containers?
• What's the live split of people working across different programming languages?
• What types of apps are deployed? Do we have hotspots (at Roche, 96% of our Connect traffic is from 2 swagger APIs)
• Can we identify potential compliance issues, like storing patient data in personal home drives, from operational metrics like user level homepod size/size changes?

What are other SCE teams doing to leverage operational data to help understand their platform and their users? Can we pool this knowledge?

Expected impact

Sharing experiences of how we use our data to optimise our platforms could benefit SCE teams across Pharma.

Prior discussions/work

I've been working on an abstract with this topic, and I'm sure there are lots of great ideas being implemented.

Would you be willing to potentially facilitate this discussion?

Maybe/No/ask me again later

[epijim]

Moved to agenda.