Deadline Cloud adds scheduling modes
AWS Deadline Cloud (a render/job scheduler) added configurable job‑scheduling modes for queues, giving teams more control over how work is prioritized and dispatched. That’s practical for studios and batch‑processing teams who need predictable slotting and different queue behaviors for bursty jobs versus steady pipelines. If you run render farms or large batch jobs on AWS, the new modes could simplify orchestration and cost trade‑offs. (x.com)
AWS announced on April 2, 2026 that Deadline Cloud now lets teams choose how cloud capacity is shared among submitted jobs, replacing the single default behavior that tended to pour all machines into the top job and delay progress on others. (aws.amazon.com) The service exposes three named modes: priority FIFO — which preserves the earlier behavior of finishing the highest‑priority, earliest‑submitted job before moving on; priority balanced — which deliberately spreads machines across multiple jobs that share the same top priority so several jobs make forward progress; and weighted balanced — which uses a calculated score to decide allocations, taking into account job priority, how long a job has waited, how many rendering tasks it contains, and recent error counts. (aws.amazon.com) A Deadline Cloud “queue” is the list where submitted jobs wait to be processed, and those jobs are executed by one or more “fleets” — groups of worker machines that run the tasks that make up each job; you set the scheduling mode when you create or update a queue in the Deadline Cloud console or via the API/CLI. (docs.aws.amazon.com) Before this change, Deadline Cloud scheduled jobs in a best‑effort priority order from highest to lowest and used submission time as a tiebreaker, which meant available workers were often assigned to the single highest‑priority job; the new modes change that dispatch logic so allocation can be balanced or weighted instead. (docs.aws.amazon.com) The weighted balanced mode’s inputs are configurable, so teams can tune the scoring to favor shorter jobs, older jobs, or penalize jobs with many recent errors, and that configurability is documented in the Deadline Cloud scheduling guide. (aws.amazon.com)
