I'm not an ops guy, but I know it was a constant source of trouble in our team and a large challenge to keep it running smoothly. For us it took 2-3 experienced engineers something like 2 years to have a stable and smoothly running production environment at scale.
If you start from scratch today things may have gotten better. You might want to look at https://github.com/overhangio/tutor, I know Régis has been hard at work making it easier to run Open edX.
If you start from scratch today things may have gotten better. You might want to look at https://github.com/overhangio/tutor, I know Régis has been hard at work making it easier to run Open edX.