Question

Your platform sends notifications through a vendor (Twilio-style) using ONE shared API key for all your customers. At 13:00 a subset of your tenants report notifications delayed or dropped. Dashboards: the vendor returns a rising share of 429s with Retry-After; your total send rate to the vendor is at the per-key limit. One large tenant kicked off a bulk campaign at 12:55 that accounts for ~80% of current sends; smaller tenants' urgent messages (password resets) are stuck behind it in the same queue and same key. How do you triage and mitigate?