Streaming Video Encoding & Packaging Pipeline  ·  View 06 of 22  ·  People and journeys

Journey — Getting a Title to Air

The operator's journey, where the worst moment is three days before anyone presses play.

Editable source SVG draw.io All views
Content operations assumed 24 staff, 1,800 versions/day Goal — Have tomorrow's premiere playable everywhere before the embargo lifts Trigger — A delivery landing, or a dated release calendar Done when — Published, gate-passed, with hours of warning if it will not make it 1 · Delivery lands 2 · Conformance ◆ moment of truth 3 · Encode & prove 4 · Publish ◆ moment of truth 5 · Watch it live What they do Books it into a lane Reads the rejection reason Chases a redelivery Watches chunk-seconds Approves the swap Rolls back if needed What the platform does Checksum & probe Conformance verdict Lane scheduling Blocking QC gate Atomic manifest swap Rollback in 120 s How it feels Confident Fine Exposed Where it hurts Rejection 3 days before air No warning a deadline will slip What answers it Verify before scheduling Verdict within the hour Chunk-seconds, not stage names Launch subset first Rollback needs no fleet Journey — Content Operations Gets a Title to Air The trough is conformance, and the fix is cheap: reject within the hour so the redelivery still fits the calendar. v 1.0 · owner Media Platform Architecture

The trough is conformance, not encoding

  • A rejection that arrives three days before air is unrecoverable: the redelivery will not fit the calendar.
  • The fix is cheap and entirely about ordering — verify integrity and conformance before scheduling any encode work, and return a verdict within the hour.
  • This is why intake verification sits ahead of analysis in view 02 rather than being folded into the first encode task.

Progress has to be legible

  • Completed chunk-seconds against total, not a stage name: an operator needs to know a deadline will slip while there is still time to change the lane or the ladder.
  • Deadline-derived priority (Phase 2) replaces the static lane field, because a static priority is wrong by the second day.

Assumption

  • Assumed 24 content-operations staff handling 1,800 title versions a day, which is why nothing in this journey can require per-title human attention by default.