NAS Disk Test - 3rd test extremely slow.

StephenB wrote:
Personally I'd have kept it turned on with those stats, but of course there is no right (or wrong) answer on this.

I know this is a difficult question, but I will ask it anyway.
In your opinion how much spin down time in hours is one spindown (and spin up) worth in degradation of the disc?
I know 30 minutes spin down time is NOT worthwhile.
I suspect that 12 hours of spin down time is worthwhile.
Somewhere in between is a crossover point.
Unless I (and others) know that value, there is little point in doing extensive analysis to work out if having the facility is worthwhile.

I've looked for studies on this topic, but the ones I've found are all from the mid-90s - and disks are very different now. So my opinion reflects my intuition more than anything else.

On the power side, your spin-down stats suggest that you'd save at most 70-75 kwh of electricity per year in direct power costs. That would save me about $20-$25. I'd take those savings if they are free, but they aren't enough for me to spend a lot of time trying to maximize them.

At +6 spin-ups per day, each drive spins up about 10,000 times over 5 years. It seems to me that the drives should last at least as long as drives running 24x7, given that you are also reducing the spinning hours over that timeframe by 15,000 hours. If the spin-ups exceeded the hours saved (e.g, 12 spin ups per day for 8 hours saved), I'd probably increase the threshold. But there's no engineering behind that, it just seems sensible to me.

EDIT- the 70-75 kwh power savings is an upper bound (real savings would be less). I based it on the difference between average power and standby power (4.5 - 4.1). It's probably more accurate to use the difference between idle power and standby power (3.3 - .04). If you do that, the power savings estimate drops to 50 kwh per year.

StephenB wrote:

I've looked for studies on this topic, but the ones I've found are all from the mid-90s - and disks are very different now. So my opinion reflects my intuition more than anything else. On the power side, your spin-down stats suggest that you'd save at most 70-75 kwh of electricity per year in direct power costs. That would save me about $20-$25. I'd take those savings if they are free, but they aren't enough for me to spend a lot of time trying to maximize them. At +6 spin-ups per day, each drive spins up about 10,000 times over 5 years. It seems to me that the drives should last at least as long as drives running 24x7, given that you are also reducing the spinning hours over that timeframe by 15,000 hours. If the spin-ups exceeded the hours saved (e.g, 12 spin ups per day for 8 hours saved), I'd probably increase the threshold. But there's no engineering behind that, it just seems sensible to me. EDIT- the 70-75 kwh power savings is an upper bound (real savings would be less). I based it on the difference between average power and standby power (4.5 - 4.1). It's probably more accurate to use the difference between idle power and standby power (3.3 - .04). If you do that, the power savings estimate drops to 50 kwh per year.

Can you re-state a case (any case) if I say "forget the cost of electricity" .. not interested in that.
I am interested in prolonging the life of the discs, and hence my operational system and minimizing the risk of downtime.
P.S. $25 in London is the cost of 1/2 a meal out, and I eat out over 50 times a year. (Not rich, just careless with my money :rofl: )

Well, as far as I can tell there are basically four things people want to accomplish with spindown.
(a) some people just want the NAS to be quiet, particularly off-hours. Perhaps they live in a studio apartment ...
(b) some people want to minimize their power use. They might be wanting to minimize their carbon footprint, or they might have severe power constraints (at least one NAS owner lives on a boat).
(c) some people want lowest cost of ownership. So they would want a crossover point that minimizes (power cost + disk replacement cost)
(d) some people want to maximize disk life. That group includes you.

I don't know of any study of reasonably modern drives that would give you a solid engineering answer to your crossover question. There was a paper published in 2010 (https://www.cs.kent.ac.uk/pubs/2011/3237/content.pdf) that cites an 1994 study that concluded "Each power state transition provides the same wear as 3.75 hours of continual operation."

That is exactly the kind of statement you need; I just don't believe its the right number for your drives - and I haven't seen any statement like that in a more recent study. WD40EFRX drives are mechanically very different from the drives that were on the market in 1994. One key difference is the bearings - fluid bearings are used today, but weren't in use back then.

My instincts are telling me that a spin down/spin up cycle is about the same as 1 hour of operation. That's why I'd leave spindown running using your settings. But that crossover point is based on my intuition, not any data - and I am not a hardware/mechanical engineer. I am a software/systems guy by training.

StephenB wrote:
My instincts are telling me that a spin down/spin up cycle is about the same as 1 hour of operation. That's why I'd leave spindown running using your settings. But that crossover point is based on my intuition, not any data - and I am not a hardware/mechanical engineer. I am a software/systems guy by training.

Thanks very much StephenB, that's what I wanted. I value your opinion and it (on this point) is probably much better than my guesswork. Cheers.