Hacker News Reader

Comment by pwinnski | Hacker News Reader

I treasure these reports, and yet every time I go to buy more hard disks, it is very difficult resist the low, low prices of those Seagates. I know they'll fail faster than the WD/HGSTs, but they're so much cheaper!

Clearly, given these numbers, Backblaze thinks the same way.

View on HN

Yev from Backblaze here -> price plays a large factor in our purchasing decisions, but you'll notice if you look at the failure rates that they're not THAT much different across the manufacturers. For home usage, you're likely going to be fine regardless of what drives you buy. Though those sweet sweet 4TB HGST drives are getting harder and harder to find!

To me, cheaper disks will always win because I expect hard disks to fail and take precautions against data loss anyway. If they fail within the warranty period it doesn't really matter where they came from. If they fail outside of it I probably still got my money's worth.

The net effect, paradoxically, is that this encourages and supports low quality. Or perhaps it redirects the resources towards those who need it most, and the average quality improves?

A disk failure is still a cost though in terms of restoring a backup, replacing the failed drive, waiting for an RMA/new drive to come, etc.. If the price delta isn't anything crazy I'd prefer to get something that I know is more reliable.

Agreed. Within $10/20, of course I'll go for the better drive. When it's $50/60 or more, I get the Seagate and hope I'm lucky.

Earlier this year I had a Seagate fail within months of purchase, putting my NAS into degraded performance mode. So I quickly ordered a replacement drive (a WD Red) and started the RMA process for the Seagate. At some point during this process I removed and reinserted the Seagate, and... its been fine since. So now I have a WD cold spare in my closet for whichever drive fails next.

I have a few RAIDZ/RAIDZ2 arrays and I've been wondering ... are let's say 8TB from WD/Seagate all the same size making them interchangeable? What if you have an array with WD drives and the Seagate drives turn out to be a few hundred/thousand bytes smaller?

The useable vdev space depends on the smallest drive in the vdev. ie, let's say you have a mirror with 1x8tb and 1x7.9tb, the vdev will be 7.9tb large. In the case of it being just a few bytes smaller, I would assume you just loose a block.

Only if all the disks when the vdev were created were the 7.9tb drives.

If you have a pool of 8tb drives and want to replace with a 7.9tb drive, you're out of luck.

I've intentionally under-allocated with array space on my drives and use the rest mostly for swap (which I don't want on ZFS anyway, mdadm raid10 is fine) and backups of /boot; all of which I can sacrifice without worry to make the data fit inside of smaller drives.

> I've intentionally under-allocated with array space on my drives

Same here - created on each drive a partition slightly smaller than the HDD's capacity (using ZFS raidz1).

After getting burned on a consecutive $3800 worth of dead-on-arrival Seagate drives, well, never again.

Just had a third well reviewed Seagate Nytro Enterprise SSD fail after a very short period of use, very low write volume. Starting to wonder whats up with Seagate.

Wish Backblaze had SSD stats.

If reliability is a concern, you could still put them in a RAID array.

(but of course this may not be economical in every situation)

Indeed, all of my disk purchases go into either my WD PR4100 NAS or the USB-connected Drobo next to it.

These types of things aren't free, but they're inexpensive enough to be worth it, IMO.

Reply on news.ycombinator.com