As SRE, strongly disagree. On Call is like hiring civil and structural engineers then holding them responsible when their poor bridge collapses under the weight of all the traffic.

Sometimes, yes, Devs get called out for stuff outside their control like infrastructure failing. However, at my job, we just had two devs that quit over on call and guess what, their service was one of worst offenders in "Opps, we pushed bug to production."

firstly, on call means supporting the entire service. not something in general I built.

secondly, many if not most of the issues that arise are part of some infrastructure automation or third party service or database. expecting me to be fluent in all of those to be useful in the hot seat is a pretty substantial investment and qualifies me to be an SRE on top of my other duties

thirdly, one major reason why my code might fail in production is that it wasn't sufficiently tested, probably because the service as a whole is basically untestable, and even if it were, building test and test infrastructure is likely not at all valued. in many places just filling in that hole would take a year.

onto to the fourth, the story is supposed to be that by operating the service, I'll be incentivized to fix automation and come up with solutions to make it more robust. I actually know how to do this, and every week I'm on call is time that I _dont_ spend doing this. furthermore, getting permission to do so is often like pulling teeth. sounds complicated. sure that would be nice, look at that when you have time in the indefinite future.

so what this often looks like from a development perspective is that I'm being paid to be a developer, I was judged based on my ability to be a developer, but at the end of the day I'm not building the service. I _am_ the service.

Give them the time and budget to build it like a bridge then. Oh wait, your competitors beat you to market by several years.

Making people work 24/7 is not conducive to good anything, thus on call is a terrible way to do things.