We rolled out API-based dock scheduling at our Dallas DC in November, wiring the YMS to the TMS with webhooks and EPCIS 2.0 events, and cut average dwell by 22 minutes per trailer. Who else is streaming milestone updates instead of batch EDI 214, using Kafka topics, Azure Event Grid, or your TMS vendor’s bus, and how did you handle carrier buy-in?
22-minute cut is strong. We got buy-in via a ‘milestone SLA’ dashboard; how’d Azure retries behave?
@steven70 > a ‘milestone SLA’ dashboard; how’d Azure retries behave? We capped Event Grid retries at 12h and gave carriers a self-serve replay endpoint (idempotent by eventId) so they could fix failed webhooks without tickets; we still let small fleets stay on 214 for 30 days to land the buy-in — do you gate SLA start on geofence+API?
We ditched 214s for Kafka + EPCIS 2.0 and got buy-in by tying detention start/stop and quick pay to carriers’ ‘arrived’/‘departed’ events. For the long tail we send an SMS magic link that posts a minimal ObjectEvent to our capture API — no TMS integration. Curious if @steven70 also scored carriers on event completeness rather than relying solely on retries.
We wired “EPCIS 2.0” into Kafka at our Joliet DC and dwell only moved after a 90s debounce + 200m geofence on “arrived” to kill ELD jitter and bad clocks. For carriers, we shipped a mock topic with canned payloads so they could certify in under an hour — tiny delay, but disputes vanished. In Dallas, did you validate “arrived” via GPS + gate sensor or trust the webhook alone?