Re: F4-220 FW 4.2.18 no HDD sleep
Posted: 28 Nov 2021, 08:17
In 41 years of running / building / programming / repairing computer systems, it is my opinion that traditional (spinning platter) hard disks seem to last longer when left "on" continuously.
Perhaps this is just me, but I've had a dozen or so disks fail, and seen many more, which were all in power-cycled or sleep applications. I think it is the thermal stress - heating / cooling causing expansion / contraction - which contributes to premature failure.
In one machine I have two 80GB SATA server disks (bought when SATA disks were just catching on) in a RAID0 stripe. It's probably comparable to a modern disk in terms of speed, so nothing special, but it was really nice back then. And it has been running (in several machines) ever since... for 20+ years. It is never shut off, and the disk temperatures are always controlled. I keep them running now as a "downloads" drive, just to see how long they will last.
I'm hopeful that the (modern, expensive) disks in my TNAS fare as well. So I set the TNAS disks to never sleep, and set the fan to Medium, and that seems to (now) keep things evenly cool under tested conditions. But I would personally like more control of the fan settings, such as "keep disks at 30°C always."
Of course, it will cost electricity to run that disk. I estimate that one (spinning platter) hard disk, active 5% of the time, will produce about 5.6W of waste heat. From an 80% efficient power supply, that means 7W from the outlet. Times 24h, times 7 days, times 52 weeks = 61kWh of electricity per year. If each kWh costs $0.10, that's $6.10 per year to operate that disk continuously. Which would be nice to reduce. Except that, either way - after a few years (power cycle) and cost of new disk, or many years (continuous) and the extra electricity - the total expense will end up about the same. That's why I choose continuous operation: longer data retention and less work for the operator, but higher electricity bill.
Best wishes.
Perhaps this is just me, but I've had a dozen or so disks fail, and seen many more, which were all in power-cycled or sleep applications. I think it is the thermal stress - heating / cooling causing expansion / contraction - which contributes to premature failure.
In one machine I have two 80GB SATA server disks (bought when SATA disks were just catching on) in a RAID0 stripe. It's probably comparable to a modern disk in terms of speed, so nothing special, but it was really nice back then. And it has been running (in several machines) ever since... for 20+ years. It is never shut off, and the disk temperatures are always controlled. I keep them running now as a "downloads" drive, just to see how long they will last.
I'm hopeful that the (modern, expensive) disks in my TNAS fare as well. So I set the TNAS disks to never sleep, and set the fan to Medium, and that seems to (now) keep things evenly cool under tested conditions. But I would personally like more control of the fan settings, such as "keep disks at 30°C always."
Of course, it will cost electricity to run that disk. I estimate that one (spinning platter) hard disk, active 5% of the time, will produce about 5.6W of waste heat. From an 80% efficient power supply, that means 7W from the outlet. Times 24h, times 7 days, times 52 weeks = 61kWh of electricity per year. If each kWh costs $0.10, that's $6.10 per year to operate that disk continuously. Which would be nice to reduce. Except that, either way - after a few years (power cycle) and cost of new disk, or many years (continuous) and the extra electricity - the total expense will end up about the same. That's why I choose continuous operation: longer data retention and less work for the operator, but higher electricity bill.
Best wishes.
