@baqui63 I guess I would if I cared. It’s such a minor system that, other than that I saw the email that it was down, no one in my organization even knew. Maybe the redundant systems are other VMs on the same host. That would be outstanding.
If they were down for two or three days, some things would start to get noticed. But what this service does for us I could probably actually do myself with a raspberry pi if I had the time and inclination. (I could use two for redundancy.)
@baqui63 I have to explain this to my customers fairly regularly. If you put both servers in the same rack it helps for availability during system patching, but it doesn’t do anything for disaster recovery when the roof caves in (which has happened to multiple customers’ server rooms).
@fibrs86 yeah… one has to balance a number of risk and cost factors when locating systems.
This conversation has gotten me thinking (again) about whether or not I should split my Cisco Hyperflex cluster between two racks (in the same room) or just leave the six boxes alone in the single cabinet where they are. Each box already has dual power circuits and the entire cabinet has dual (10gE) network connections. The various (known) roof leaks are 15+ feet away and it would cost big bucks to move things more than about 5m apart, so I’ve come to the conclusion (yet again) that there is no good reason to separate them.
Plus, the additional risk has been communicated up the chain and the VP of Finance has decided that the risk is not great enough to warrant putting another cluster on the other side of campus.