I concur and agree in the most part But to anyone who is on the edge of being tempted, you need to factor in failure of parts and the replacement times. So check your onsite SLA, storage and replacement part stock holding. Once you do begin to 'worry', you're into dual PSU, multiple hot swap and or a second failover replica. I have spec'd whole racks, multi dc failover for clients and have my own relics here on the shelf. I've not been happier since the day I got rid of my own purchased tin from my production stack. If I was doing it again, I'd not rely on just one of my own boxes, at worst next day on site = 12hrs downtime. Rented tin or vps, it's not the cheapest, but the differential isn't worth the headache when it fails, and it will. You can relax when you have HA in place. Even then, a digger cuts your fibre, an errant fire destroys a key junction or your dc runs out of fuel in a flood... But you can easily overspend for a once in a blue moon situation.