Seagate Exos X10 SAS - 10 TB - $95

[DaveInTexas]

Anyone have experience (good or bad) purchasing used enterprise drives from Unix Surplus? I've had good luck buying servers and mobos from them, but I have not (until now) contemplated purchasing drives from them.

This deal of < $100 for a 10 TB refurbished enterprise drive recently caught my attention: (Server Drive) Seagate 3.5in 10TB SAS HDD Enterprise Exos 7200RPM 12GBs SAS-3 256Mb v6 512e

It's a Seagate Exos X10 SAS drive 10 TB. This drive is essentially an Enterprise Capacity gen 6 SAS drive re-branded and made colorful by Seagate's marketing department. The EC drives are some great workhorses. I still have a few. I've never bought any (so far) since Seagate pretended to release a new product line when they re-badged some of these drives as part of their Exos line. The specs are identical, but I don't know if the internals are or not.

Just to be clear... these are nearline drives, so I'm not sure how much of their SAS3 bandwidth you can truly saturate, but still seems like a decent deal to me, all things considered (unless they have very high hours or aren't actually refurbished as claimed). I'm also not sure who refurbished them. That could mean Uncle Johnny wiped the dust off them or it could mean a true refurb by Seagate (not likely, me thinks, but just sayin').

 

[Samir]

I think there's a couple of recent posts about then if you do some searches (I can't remember what the thread topics were).

 

[DaveInTexas]

FWIW to anyone reading this thread, the SKU on Unix Surplus website maps to the Seagate Enterprise Capacity 3.5 v6 model, circa 2017. The Exos X10 is circa 2016. The photo shows the Exos, but it may be a stock photo. Regardless, either the photo is wrong or the model # / SKU is wrong.

ARR (Annual Rate of Return) on the Exos X10 model is 1.32%. I don't have any return ratio data on the EC v6, but theoretically they ought to be nearly identical. Reliability is near the median of enterprise HDDs. Not bad, but not spectacular either.

 

[Cruzader]

Anyone have experience (good or bad) purchasing used enterprise drives from Unix Surplus?

Quite alot of bad experiences posted about them when it comes to drives.

Ive bought some servers/shelves/switches type stuff from them but would not have bought drives from them.

 

[ocfguy]

FWIW to anyone reading this thread, the SKU on Unix Surplus website maps to the Seagate Enterprise Capacity 3.5 v6 model, circa 2017. The Exos X10 is circa 2016. The photo shows the Exos, but it may be a stock photo. Regardless, either the photo is wrong or the model # / SKU is wrong.

ARR (Annual Rate of Return) on the Exos X10 model is 1.32%. I don't have any return ratio data on the EC v6, but theoretically they ought to be nearly identical. Reliability is near the median of enterprise HDDs. Not bad, but not spectacular either.

OT but do you have any ARR data on the X16 16/14TB and the X18 18TB SATA? Backblaze seems to be showing some higher than average failure rates...

 

[DaveInTexas]

OT but do you have any ARR data on the X16 16/14TB and the X18 18TB SATA? Backblaze seems to be showing some higher than average failure rates...

On the X16 14TB and 16TB flavors, yes. X18 18TB, no. I also have data on the X16 12TB model, FWIW. Interesting to see the rise as capacities increase, which incidentally is a common problem, especially for a few years after introduction.

manu/gen	capacity	ARR	model	# of samples	avg sample age (months)
Exos X16	12 TB	0.63%	ST12000NM001G	12,171	14
Exos X16	14 TB	1.02%	ST14000NM001G	10,738	11
Exos X16	16 TB	1.16%	ST16000NM001G	10,861	8

 

[DaveInTexas]

btw, those ARR numbers are actually decent, relatively speaking.

 

[DaveInTexas]

If a low ARR is important to you, and cost is not, I would suggest considering something like 16TB WD Ultrastar DC HC550 (SED - Self Encrypting Drive) model # WUH721816ALE6L1 with 0.14% ARR (grain of salt alert: 1,767 samples w/avg 5 months), released the same year (2019).

 

[ocfguy]

If a low ARR is important to you, and cost is not, I would suggest considering something like 16TB WD Ultrastar DC HC550 (SED - Self Encrypting Drive) model # WUH721816ALE6L1 with 0.14% ARR (grain of salt alert: 1,767 samples w/avg 5 months), released the same year (2019).

Do you have data on HC550 non-SED 16TB (WUH721816ALE6L4) and X18 16TB SATA by chance?

 

[DaveInTexas]

Do you have data on HC550 non-SED 16TB (WUH721816ALE6L4) and X18 16TB SATA by chance?

WUH721816ALE6L4, yes:

manu/gen	capacity	ARR	model	# of samples	avg sample age (months)
Western Digital Ultrastar DC HC550 (SE)	16 TB	0.23%	WUH721816ALE6L4	1,767	5

DOR (Date of Release) 2019.

specs: https://documents.westerndigital.co...t-manual-ultrastar-dc-hc550-sata-oem-spec.pdf

Unfortunately, I don't have any field reliability data on the X18 series. Just the official AFR of 0.35%. Personally, I would have more confidence in the URE estimate of 1 per 10^15 bits than the AFR, when it comes to predicting risk of data loss, though they are both of arguably limited usefulness. Another perceived decent gauge of reliability may be workload on the X18 (550 TB/year) and its 5-year warranty. I like to think the HDD manufacturers have considered their potential liability risk before they announce these drives' workload and warranty ratings. I think these days that might be one of the best gauges of 'presumed' longevity for a product that hasn't been out very long (2 years or so), until us geeks are able to suss out real-world return data.

 

[ocfguy]

WUH721816ALE6L4, yes:

manu/gen	capacity	ARR	model	# of samples	avg sample age (months)
Western Digital Ultrastar DC HC550 (SE)	16 TB	0.23%	WUH721816ALE6L4	1,767	5

DOR (Date of Release) 2019.

specs: https://documents.westerndigital.co...t-manual-ultrastar-dc-hc550-sata-oem-spec.pdf

Unfortunately, I don't have any field reliability data on the X18 series. Just the official AFR of 0.35%. Personally, I would have more confidence in the URE estimate of 1 per 10^15 bits than the AFR, when it comes to predicting risk of data loss, though they are both of arguably limited usefulness. Another perceived decent gauge of reliability may be workload on the X18 (550 TB/year) and its 5-year warranty. I like to think the HDD manufacturers have considered their potential liability risk before they announce these drives' workload and warranty ratings. I think these days that might be one of the best gauges of 'presumed' longevity for a product that hasn't been out very long (2 years or so), until us geeks are able to suss out real-world return data.

Thanks! Do you happen to work for a hyperscaler lol

 

[DaveInTexas]

Thanks! Do you happen to work for a hyperscaler lol

he he... no.... I've been collecting this type of data on HDD and SDD reliability for awhile. On my personal 'to-do' list is to publish this information as some of it is very difficult to find, and frankly it's a huge pain for anyone to try and dig up this type of info at the moment it's relevant, since most of us aren't purchasing disks on a routine basis.

I've managed to scrape the data I have from about a dozen or so sources over the years. Mostly international data sets. As I believe you mentioned in a previous post, Backblaze is a good resource if they happen to have published info on a drive relevant to someone's interest. They're one of the few U.S. based sources of info the HDD companies haven't tried to squash.