What are industry best practices for this stuff? Do you want a separate fabric for dev and prod? Just separate zones?
My company doesn’t connect any fabrics (1 switch=1fabric) and it seems inefficient and a giant pain in the ass, as we move hosts around like we’re getting paid for it. What do all of you do?
I am not a SAN admin but work closely with them. So take this with a grain of salt.
Best practice is always going to be to split things into as many failure domains as possible. The main argument being how would you test upgrades to the switch firmware without potentially affecting production.
But my personal experience says that assuming you have a typical A/B fabric that is probably enough to handle those sorts of problems, especially if you have director class switches where you have another supervisor to fail back to.
I’ve personally seen shared dev/prod switches for reasonably large companies (several switches with ~150 ports lit on each switch), and there were never any issues.
If you want to keep a little separation between dev and prod keep those on different VSANs which will force you to keep the zones separated.
Depending on how strict change management is for your org keep in mind that tangling dev+prod might make your life worse in other ways. i.e. you can probably do switch firmware updates/zoning changes/troubleshooting in dev during work hours but as soon as you connect those environments together you may have to do all of that on nights and weekends.
I’d like to see separate fabrics for prod/dev. We have a few switches that are platform specific though that would be really nice to have integrated into the general prod environment. As those platforms are absorbed into our main platform that environment needs to be repurposed anyways and it would be nice to just do it logically without having to recable everything when something needs to be changed.
