idea: Portal Performance "Chaos Engineering"

[allaway]

Intentionally stress-test the portal with realistic but extreme scenarios. What happens when someone tries to download 50 datasets at once? When 100 people search simultaneously? Create a report documenting breaking points, surprising behaviors, and recommendations. This kind of testing rarely happens in academic infrastructure but is critical for reliability.

(brainstorming help from Claude)