> Admittedly, that's probably less of a thing now we have redundancy and continuous deployment practises; the concept of a late night / weekend change window is diminishing in my personal experience. I'm intrigued if that's true across the tech industry.
Odd, my experience is exactly the opposite. Increase complexity breads increasing numbers of potential problems, which causes partial outages at unpredictable times and for unpredictable reasons (or reasons so complex they are perceived as unpredictable).
Now are these "user impacting?" Often times, no. But they will become user impacting if left for too long (e.g. new instances aren't able to connect to a database, but existing ones are connected: classic DNS problems). So people are still waking up in the middle of the night to fix stuff when the alerts start streaming in.
The only thing that has REALLY changed is that now you can often fix these issues from your bedroom in your PJs. Since the AWS, Azure, Rackspace, etc console doesn't care where you are or what you're wearing.
But companies are getting no better at actually having around the clock staff. When they used to out-source a dedicated server the data centre would employ 24/7 staff on real shifts, Amazon/Rackspace/Google/etc do the same, but for the software side of things you rarely see it (never?). Most companies just expect employees to be on-call 24/7 as part of their salary (unpaid overtime essentially).
Odd, my experience is exactly the opposite. Increase complexity breads increasing numbers of potential problems, which causes partial outages at unpredictable times and for unpredictable reasons (or reasons so complex they are perceived as unpredictable).
Now are these "user impacting?" Often times, no. But they will become user impacting if left for too long (e.g. new instances aren't able to connect to a database, but existing ones are connected: classic DNS problems). So people are still waking up in the middle of the night to fix stuff when the alerts start streaming in.
The only thing that has REALLY changed is that now you can often fix these issues from your bedroom in your PJs. Since the AWS, Azure, Rackspace, etc console doesn't care where you are or what you're wearing.
But companies are getting no better at actually having around the clock staff. When they used to out-source a dedicated server the data centre would employ 24/7 staff on real shifts, Amazon/Rackspace/Google/etc do the same, but for the software side of things you rarely see it (never?). Most companies just expect employees to be on-call 24/7 as part of their salary (unpaid overtime essentially).