Included in every plan:

Jobs fire at the top of each minute according to their cron expression. If a schedule says */5 * * * *, the job runs at :00, :05, :10, and so on. No polling delay, no drift.

Your plan's concurrency limit is the maximum number of jobs that can execute at the same time across all of your schedules. When every slot is in use, new executions queue until a slot frees up.

Each schedule can add a random splay delay (up to 45 seconds) so that jobs don't all fire at :00 simultaneously. A wider splay reduces peak contention at the cost of less precise start times, which is ideal when you have many schedules sharing limited concurrency slots.

Jobs queue when all concurrency slots are occupied. Two factors increase queuing: longer-running jobs that hold slots longer, and lower concurrency limits that provide fewer slots. If queuing is frequent, consider a higher-concurrency plan, shorter job timeouts, or adding splay to stagger start times.