[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ale] 2017 Backblaze HDD Failure Report

On 2018-02-02 06:49, Derek Atkins wrote:
> Alex Carver via Ale <ale at ale.org> writes:
>>> When I looked at the data, I saw that WD-Reds were not so great and the
>>> HTSG 4TB were the best for failures.  There are a number of different
>>> charts in the article. Best to carefully check the captions.
>> The HGST are helium filled drives and cost significantly more than the
>> WDC Reds.  The others that are on the same price point seem to also have
>> the same longevity.
>> One person did the life expectancy calculations for the drives here:
>> https://medium.com/@simonerni/dissecting-backblaze-hard-drive-stats-2017-17a90089a2e8
> I happen to love HGST drives, and I'm willing to spend a LITTLE more to
> get them.  The 4TB drives are ~$130 each on newegg.  I wish there were
> more data on the HGST 6TB drives, as they are under $200 each and
> currently the best $/TB of the HGST series.
> I currently have 6 4G HGST drives in my FreeNAS server (RAIDZ2).  I'm
> due to add new space soon.  ;)
> -derek

The problem with helium drives is that helium leaks out (which is why
helium is used for leak testing vacuum chambers).  The drive is relying
on the helium for cooling (transferring heat from the platters to the
case) and to keep the heads above the platter at a gap smaller than a
typical non-helium drive.  When the helium finally leaks out you're left
with a drive that is overheating and not far away from a full head crash.[1]

No matter what anyone says about their drives having "solved" the helium
leak problem, I will guarantee you that *nobody* has fully solved the
helium leak problem because helium *will* leak out.  I have vacuum
chambers at work that are supposed to reach pressures equivalent of
interstellar space and I can *still* find helium inside the chamber when
I spray it with helium from the outside.

[1] The loss of cooling will cause the platters to expand, narrowing the
already narrow gap between the platter and the head.  The loss of helium
also means there's not enough gas to get under the heads.  Nitrogen and
oxygen are too big to fit under the head reliably to provide the air
bearing so the head will crash.