Critical services have their own terminology for this that lessens the ability to weasel around phrases like "24/7", "round-the-clock", "Mission critical", etc.
It's the "Nines" system and is common in legal contracts for "High availability" services.
Calculate "Nines" by taking the percentage of uptime over total time. For example, if a server was up for only 9 out of 10 seconds, that's 90% or "One nine".
The common "Five nines" is 99.999% which still allows 6 seconds of downtime per week.
To get to less than 1 second per week, you need to guarantee "Six nines" or more -- which is a herculean task in the long term. (Even more-so if you don't use the standard "Unplanned" weasel word.)
For example, Google and Microsoft typically only promise "3 nines" (and frequently fail even that), which is a whopping 10 minutes down per week.