jimmysmitty :
The Backblaze data is also useless because most of the HGST drives are their enterprise class drives and all the Seagates are consumer drives which invalidate the information because a consumer class drives is not meant to be used in a 24x7 high power, high heat and high stress environment
Anyone who uses the backblaze data to support an argument about reliability of consumer drives ... essentially disqualified themselves as being qualified to participate in the discussion.
The obvious question is, if Seagate Drives are so bad, why do they account for the largest number of drives in use at Backblaze ? And the answer is that, for them, it's more profitable to use inexpensive consumer drives than pay for drives that are designed for the environment they serve.
Why do consumer drives fail early in a sever environment ? When a drive is designed for a consumer environment, an important feature is "head parking".... this "parks" the arm / head away from the platter much like a tone arm / needle is returned to a parked position on a phonograph. With typical home and building construction, floors vibrate ... people bump desks as they are walking by .... PC cases get hit by elbows. With the heads "parked" no damage occurs.
Consumer drives are rated for between 250 and 500k cycles, more than enough for the anticipates life of a consumer drive. However, a server drive has I/O several orders of magnitude above a consumer drive ... that head / arm almost never stops moving. A server can burn thru 250 - 500k cycles in a matter of months.
If we look at actual data for consumer drives being used in a consumer environment, we see this combining the data for the last two 6 month study periods:
HGST 0,70%
Seagate 0,64%
Western 0,95%
Toshiba 1,05%
Those numbers represent the number of failures which occurred with drives between 6 and 12 months of usage averaged over 12 months. And while the numbers are what they, I don't look at them that much ... because like every other component you put in a PC, just about every manufacturer has made some astounding products and also has produced some duds. In storagereview.com's storage reliability survey, what can we make of the following:
Most Reliable Drive Model of all time: Made by Seagate
Least Reliable Drive Model of all time: Made by Seagate
Where the data is reliable, is in determining what specific models fare poorly
These broke 2% in the last period
2,17% WD Black WD3003FZEX
4,32% WD Black WD4003FZEX
3,59% Toshiba DT01ACA300
2,88% Toshiba DT01ACA200
2,88% Toshiba DT01ACA200 2 To
2,39% Toshiba PA4291E-1HJ0 1 To
2,17% WD Black WD3003FZEX
And in the one before that ....
4,90% Toshiba 3 To DT01ACA300
2,86% WD RE 4 To WD4000FYYZ
2,33% WD Blue 250 Go WD2500AAKX
2,23% WD Black 4 To WD4003FZEX
2,20% WD Red 750 Go WD7500BFCX
It's not as if the absence of a drive on the list is necessarily a good thing.... if sales don't reach a minimum number as to be statistically significant, the drive won't be listed.
Favoring a drive by brand name can be just as foolish as picking a PSU by brand name. Just cause the Corsair AX 860i is a great PSU, doesn't make the CX 600 a great PSU. Best you can do is pick a drive model that has consistently had an acceptable failure rate .... and recognize that, all things being equal, increased performance may mean lesser expected life.
Facebook
Twitter
Instagram