i like this answer, because it's actually true. Spindowns are only permitable when the electricity bill is too high, and the system is idle for at least 6+ hours.

Bạn đang xem:

TL;DR things like to lớn stay spinning, start/stop is where the damage và failure happens.

Durability = ức chế them as little as possible.

Meaning:

Minimize running time AND start/stop count... which are opposites, but that's the compromise.

Keep them at a constant temperature AND cthất bại to ambient... also opposites, but spikes of hot-cool-hot are as bad as cool-hot-cool.

You only want to lớn keep your HDDs spinning all the time if you're accessing them at least once ever "MTBF / max load cycles", which usually is in the order of "100k hours / 500k load cycles" or about one stop every 12 minutes. Anything more & you're wearing your drives more by spinning them than by stopping them. Unless you keep accessing them in 6 or 11 minute intervals, I'd recommkết thúc parking after 5 or 10 minutes, so they stay running all the time you're using them, & stopped when you aren't.

Performance is much simpler: SSD.

Xem thêm: Hàm Tìm Số Trùng Nhau Trong Excel 2003, 2007, 2010,2013, 2016

cấp độ 2
· 2y · edited 2y
24TB
Really? My Seagate drives would get really hot in the summer if I kept them spinning. One died at 65°C, so I turned it off for all my other drives, except for SSDs.

level 1
· 2y
Never enough TB
Yes, 0.

It doesn't matter. You'll find it very tough lớn hurt a hard drive sầu, due lớn the power settings.

As an industry insider, having worked in many different companies in hard drive development, generally from a reliability standpoint we don't care if you keep the drive sầu spinning or if you don't. The motor spins on a thin layer of fluid, và for all practical purposes, we simply don't have sầu any problems up to lớn five years (which is the life scale of the motor testing we do) và probably if we tested them, a lot of drives would live a lot longer than this. I recently pulled out a decade old 2.5" drive sầu, spun it up, & it operated just fine. I've sầu had some of my data center customers at big cloud place regularly beat on drives 7 years old, & now they are replacing them because they are too small, not because they are failing.

Load/Unloads on the ramp really isn't a factor. We regularly thử nghiệm khổng lồ a million load/unload cycles, và we don't see a problem. That's like parking và unparking for 10 days straight every second. We've sầu had a couple customer have some accidental power settings that aggressively parked on the ramp in Linux that caused problems, but that was years ago, và in the millions upon milion of cycles, and we've sầu improved ramp material since then. I haven't heard of anything like this recently, và nothing under Windows.

As a general rule, I bởi vì like the drives to be flipped on once in a while. We have sầu background operations to lớn check the validity of data written, and if we think something looks wrong, we rewrite the data. (Data on the disk has ECC on it, so we can tell if something is going wrong.)

It is possible khổng lồ eventually kill a hard drive by just writing so much information that the head fails. On the enterprise drives, we talk about 550 TB/Yr over 5 years as our spec. However, unlượt thích NAND, we don't have a clear wear out mechanism. Yes, you may find one head fails, but even on a client drive sầu, the rest may live sầu for a crazy long time. We are developing techniques where if a head fails, we'll give the ability to reformat the drives at a smaller form size, và we think it should last a long time.

For those that suggest keeping things cool, yes, almost all HDD makers at one time or other have suggested cooler is better (well maybe not absolute zero). Generally, the closer khổng lồ the top end of the spec of 60C, the more you push the drive. I lượt thích to lớn tell most of my customers that my experience says that keeping the drive sầu under 50C is something I would bởi in my own machine. However, we give you a spec of 60C, và all HDD makers work khổng lồ make their drive sầu work well even in this environment.

Somebody remarked on MTBF, but generally we talk about AFR now. A 2.5M MTBF is a .35% AFR. This means over 5 years, you'd thua thảm roughly 2% of your hard drives. And we vì chưng run thousands of drives for thousands of hours to lớn thử nghiệm for this, but HDD makers kiểm tra at more moderate temperature and at moderate workloads, not necessarily the extremes of our spec. Then rethành viên HDDs are shock sensitive sầu (static shock and physical shock) not only installing the drive, but all along the supply chain. I've seen very well executed big cloud guys get lower than .35% AFR, but it is not uncommon for many people lớn struggle to lớn get lớn .35% & 1-2% is not totally uncomtháng (see Backblaze data).

Life is also statistical variation, & all HDD makers will have sầu good runs và bad runs, just lượt thích get a hot roll on dice at a craps table. Eventually, everybody reverts to a mean. If any of the HDD makers truly were worse than the other, they'd thua trận all their customers because poor chất lượng isn't worth it.