Please explain

>>105670826 (OP)
No.

>>105670826 (OP)
it has 2TB more retard.

>>105670826 (OP)
The 24TB drives have defective platters/reduced lifespan.

If I had to guess, the 26TB ones use more expensive platters simply for 2 extra TB

>>105670857
lifespan starts to shorten significantly above 10TB, it's usually recommended to use large size hdd for cold-ish storage with few read/write.

>>105670857
>>105670875
cap

>>105670826 (OP)
The 26TB model is the world's first 11 platter CMR HDD I believe.

WD have 32TB drives but they won't sell them to me

Economies of scale.

>>105670875
>lifespan starts to shorten significantly above 10TB

any source for this? I have 4 year old 12tb drives that run just fine.

>>105671353
backblaze hdd annual report, you can see 10TB hdd having a massive spike in failures and hdd of bigger size while being newer started to fail sooner than 8TB hdd.
>I have 4 year old 12tb drives that run just fine.
you should not really worry, all hdd should last their warranty period just fine on average

>>105671398
The annual report also shows no issues on 16TB drives

You're memeing.

Certain drive sizes have historically been plagued with issues, but it's not a cut and dry once you go over 10TB suddenly issues happen.

See pic related, Toshiba 16TB drives, 40000 of them running for an average of 20 months with less than 1% failure rate.
Or the 14TB Toshiba drives, 37,000 of them running for an average of 52 months 1.4% failure rate.

The idea that anything over 10TB has a "massive spike in failures" is just moronic and ignore the ACTUAL statistical data you're claiming backs up your opinion.

>>105671398
>>105671452
HGST 8TB and 10TB has some models which use lots of platters but without helium filling, which results in the drives running at 70C without active airflow. Those will die very easily, but they are known duds. So are basically any drive with Seagate written on them, which you can see on the stats.

>>105671519
> any drive with Seagate written on them, which you can see on the stats.

>Seagate 16TB
>33,817 drives
>average running time of 30 months
>1.4% failure rate

>Toshiba 16TB
>40,273 drives
>average running time of 20 months
>0.95% failure rate

Riiight well ACTUAL statistics prove that wrong.


Again, individual models, yes. ALL Seagate drives? No.

Like we can ALL see the data there, you're not gonna be able to hide behind LE SEGATE BAD maymay.

>>105671519
>So are basically any drive with Seagate written on them, which you can see on the stats.
most of the Seagates on the chart above your post seem to have pretty low failure rates though. I haven't checked but I'm willing to bet the few that have high failure rates here are BarraCuda slop

>>105671519
Based purely on the data from BackBlaze, Seagate is statistically the worst, but not by an amount the average consumer would ever notice.

>HGST
>1.07%

>Seagate
>1.84%

>Toshiba
>1.08%

>Western Digital
>0.52%

However since Western Digital and HGST are essentially the same company I think it's fair to combine their data giving a combined average of 0.90%

>>105671452
Why is seagate so ass

>>105672998
See >>105671763
They're really not when you account for all the factors and all the drives.

Why do you need so much space for? Is ever movie on your media server 4k? Are you archiving 8k vr porn

>>105675635
>Is ever movie on your media server 4k
only the ones released in 4k

>>105670826 (OP)
Jews, Chinks, Jeets, Arabs and Yteys.

>>105677345
>omg, world leaders visit one each other and give themselves handshakes
>that means they are friends and share ideologies!
Why are /pol/tards so retarded and childish?
Also, why do you keep.spamming that pic, you sad NPC?

>>105670826 (OP)
I have 5x hc550 18tbs, they're good and pretty quiet.
I can recommend wd ultrastars, but those prices per tb are awful on the 24/26 tb models.

>>105671398
The Seagate 14TB hasn't had any failures in nearly a year of service.

>>105671763
>However since Western Digital and HGST are essentially the same company I think it's fair to combine their data giving a combined average of 0.90%
How did you come up with a number closer to the HGST rate when Backblaze has more of a single WD model than all HGST models combined?

>>105677796
Because I was doing weighted averages and including extra failure data from the backblaze post on their website, not just the data from that single image


So I was including drive days, when the drives failed, etc.

>>105671353
Wow dude 4 years daaaaamn. I have an old ass 1TB drive that's at least 15 years old now, I don't remember when I got it.

>>105678083
How many power on hours does it have though

I've got drives with around 10 years of power on hours.

>>105670826 (OP)
Because it's not priced for someone buying an individual drive.
Sure if you're just buying one, it's insane to pay that much for 2 extra terabytes, but when you're buying 2000 drives it suddenly becomes worth it to pay that much extra for 52 petabytes of storage in the exact same physical space as 48 petabytes using the cheaper drives.

>>105677711
>red*it ty*ing chink
>retarded gaslighting and self-projectting

every time

>>105678399
Yeah

>>105671519
No HGST drive that is 8-10TB is without helium.
Now there are some pre-helium HGST designs that WD has shoehorned more/higher density platters, but this is well after the acquisition and no enterprise is using those drives as it really doesn't make sense from a price per TB perspective.

>>105671763
Is that annual failure rate? Also, double the failure rate (0.9% vs. 1.84%) seems pretty significant in context.

>>105680602
Yes, that's annualized.

>>105680602
Significant for someone buying dozens or hundreds of drives, not particularly significant if you're a consumer buying a handful of drives.

AFR under 2% is generally considered great for consumer drives.

>>105670875
I don't remember there being a source for this concerning my 14tb toshiba mg drives

>>105680733
It's just retards doing the "old good, new bad"
Nevermind 10TB+ HDDs have existed for almost a decade.

>>105680720
Well, if you buy three drives with an annual failure rate of 1.84%, and own them for five years, you're at a 24.3% chance of at least one of the drives failing.
With the 0.52% chance, it's 7.5%.

p(1+fail) = 1 - ((1 - p(annual))^(drives * years))

>>105680800
Your math is off

>>105680820
You or your AI model forgot that there are three drives in this hypothetical. I took that number as what I thought was a reasonable interpretation of "a handful of drives".
Take that 8.74% it calculated across three drives, and you'll get the same result I did (within rounding).

>>105680800
>>105680820
>>105680865
Also, if we filter out the drives with less than 5000 samples and only use the data from the larger drive sets we get;

Toshiba
> 1.07%
Seagate
> 1.67%
Western Digital/HGST
> 1.17%

Which gives a 5 year % chance of dying of;

Toshiba
> 5.2%
Seagate
> 8.03%
Western Digital/HGST
> 5.69%

And if you want to account for 3 drives with any one of them failing then you get;

Toshiba
> 14.81%
Seagate
> 22.12%
Western Digital/HGST
> 16.15%


And if you want to go back and include all the drives (even models with just a few hundred samples) then it's;

Toshiba
> 15.12%
Seagate
> 23.93%
Western Digital/HGST
> 12.68%


No matter what your 7.5% number is just wrong.

>>105670826 (OP)

Look up economies of scale, doofus.

>>105680904
>>105680865
It should also be noted, these are datacenter conditions, not something you'd run into at home.

Expect that failure rate to be excessively high for home use.

>>105680904
>>105680934
So you've added extra conditions to the data, and then actually thrown out all of the data before we even get started. How can I possibly engage with that?
You're also combining HGST and WD even though a consumer could select for WD if drive survivability was their goal.
You made a minor error by over-reliance on an LLM. It's no big deal.

>>105680974
I already told you I combined the two earlier, not my fault you're not paying attention or being are purposefully deceitful to boost western digital because you own one of their drives.
>>105671763
>since Western Digital and HGST are essentially the same company I think it's fair to combine their data giving a combined average of 0.90%


If I were buying a drive today, I'd buy Toshiba based on the data and if you would buy Western digital then you're shit at reading data and I have to wonder what you're even doing here.

>>105680993
I own one WD and one SeaGate. Why would I try to boost the perceived value of a drive just because I own one anyway? It's not like it's an investment commodity or stocks.
I read your point about combining the figures. As you have seen I did not combine the figures, and I explained why in my previous post.

>>105680904
>>105680934
>>105680974
Kinda moot to really talk about drive failures with data from HDDs that have long since been out of production.
Once HDD platforms are out of general production, any lower capacity drives other than the top-end, end up as frankensteins as capacities may be moved across platforms.

If you bought a 12TB Seagate for example when they where first released you will probably got a Mobula, a 12TB today maybe based on Evans platform.

>>105681038
>and I explained why in my previous post.
Quote your explanation for me, I'll wait.

>>105681056
>even though a consumer could select for WD if drive survivability was their goal
I'm talking specifically about WD rather than your combined WD/HGST when I talk about selecting for WD here.

>>105681061
And you talk about ME ignoring data?

HGST IS Western Digital.

The reason Backblaze no longer purchases HGST is because they're now selling under the Western Digital brand.

Combining them is the sane thing to do with that dataset.