Most desks promise faster service for urgent work. A Highest priority incident might get one hour, around the clock, while ordinary requests get eight working hours. Each tier is a goal with its own JQL, target and calendar.
In a hosted SLA, the tiers are goals inside one SLA, checked top to bottom, and the first goal whose JQL matches a request applies. Order matters: if a broad goal such as priority IN (High, Highest) sits above priority = Highest, the urgent goal never applies. This practice site stores each tier as its own SLA, so you see each clock separately; keep their JQL from overlapping and every request still gets exactly one clock per promise.
Write tier JQL so the tiers cover every request once. priority = Highest and priority != Highest do that for priority, because every work item has one. Check a request created outside working hours, since that is where a 24/7 tier and a working-hours tier disagree most.
Give urgent work its own goal and keep tiers from overlapping.