Backblaze blog: What Hard Drive Should I Buy?
- Thread starter DSC
- Start date
Dont take those graphs at face value
Dispelling Backblaze's HDD Reliability Myth - The Real Story Covered
http://www.tweaktown.com/articles/6...bility-myth-the-real-story-covered/index.html
Dispelling Backblaze's HDD Reliability Myth - The Real Story Covered
http://www.tweaktown.com/articles/6...bility-myth-the-real-story-covered/index.html
It's just one data source, even though they may have quite a few drives (I dunno the total number of units they covered), so one should perhaps not draw too specific conclusions from this. A wider distribution of makes, models and running conditions would give more reliable figures. Google (semi?)periodically release HDD reliability studies, and failure rates are typically much lower than at least the seagate figures from what I recall.
I used to buy quantum exclusively until they got bought out, and then I used to buy hitachi exclusively due to what was then unique features to them (originally developed by IBM) such as parking head-ramps and low spin-up currents and such, but as these features became ubiqutous I then switched to western digital, and I've been satisfied with them since.
It's my belief you can pick any make and have pretty much equal chance of being satisfied or disappointed with any of them. Most of the horror stories you hear about reliablity problems and so on are completely anecdotal and random in nature, even those purporting to show a pattern from some manufacturer's offerings.
Some exceptions exist, such as certain seagate drives in the past, or certain IBM drives before that. Those are outlier anomalies. It is my belief there's no major differences between the manufacturers, and if there were they'd go out of business like all their competitors have done in the past.
So...pick something that looks good to you, and good luck.
I used to buy quantum exclusively until they got bought out, and then I used to buy hitachi exclusively due to what was then unique features to them (originally developed by IBM) such as parking head-ramps and low spin-up currents and such, but as these features became ubiqutous I then switched to western digital, and I've been satisfied with them since.
It's my belief you can pick any make and have pretty much equal chance of being satisfied or disappointed with any of them. Most of the horror stories you hear about reliablity problems and so on are completely anecdotal and random in nature, even those purporting to show a pattern from some manufacturer's offerings.
Some exceptions exist, such as certain seagate drives in the past, or certain IBM drives before that. Those are outlier anomalies. It is my belief there's no major differences between the manufacturers, and if there were they'd go out of business like all their competitors have done in the past.
So...pick something that looks good to you, and good luck.
It's basically at the start of the article. Out of those 3 manufacturers, total was about 30,000 drives. That's a fair pool of data (pun?) to draw statistics from.
itsmydamnation
Veteran
is it really that different, they effectively provide low speed trickle backup. if anything there drives will see lighter usage for the amount of consumed space then a regular desktop. Assuming they do patrols i bet that makes up the vast majority of reads ( out side of the parity calculation).
Their data can't be applied to home user at all.
I could do the same study using my office customers (5+ years, pool of 150+ drivers) and report that WD is the looser due to me purchasing large batch of WD 3200AAKS drives from which 80% is already DEAD! On the other hand slow and old Samsung SpinPoint 160 wins with only one failure and dozen of drives still going strong in their 7th year.
So one could think that WD is bad and Samsung is great but that's not the case as I have plenty of WD 500GB and WD 250GB drives doing great with no failures at all so far.
In my opinion best way to judge the drive is to look at default manufacturer warranty length and just scan web for any major issues reported from other users.
I use WD, Samsung and Seagate in my computers and to be honest all are good enough. Had soft spot for Sammys F1 before and loved WD Black series, but to be honest now Seagate cleared their act after Barracudas 7200.11 and I even have one new Barracude 3TB in my main PC at home
I could do the same study using my office customers (5+ years, pool of 150+ drivers) and report that WD is the looser due to me purchasing large batch of WD 3200AAKS drives from which 80% is already DEAD! On the other hand slow and old Samsung SpinPoint 160 wins with only one failure and dozen of drives still going strong in their 7th year.
So one could think that WD is bad and Samsung is great but that's not the case as I have plenty of WD 500GB and WD 250GB drives doing great with no failures at all so far.
In my opinion best way to judge the drive is to look at default manufacturer warranty length and just scan web for any major issues reported from other users.
I use WD, Samsung and Seagate in my computers and to be honest all are good enough. Had soft spot for Sammys F1 before and loved WD Black series, but to be honest now Seagate cleared their act after Barracudas 7200.11 and I even have one new Barracude 3TB in my main PC at home
Mine is not bad, I mean it's louder on power on/off than my Sammys F1 but not massively. During normal operation it is quiet, but to be fair my whole computer have one dominating noise nothing else can overcome and that is R9 290X CF mining
Get yourself two of these with stock coolers and you will not hear your drives at all Nice I have a Hitachi in my machine
30,000 samples no good enuf data!
So whats good enuf data?
I got a samsung SSD recently even though I didnt want to due to my last (non ssd) samsung HD failing (I dont want to reward failure) but ppl on this forum were saying a single data point was not good enuf, now theyre saying 30,000 samples isnt good enuf, so how many billion do you want
30,000 samples no good enuf data!
So whats good enuf data?
I got a samsung SSD recently even though I didnt want to due to my last (non ssd) samsung HD failing (I dont want to reward failure) but ppl on this forum were saying a single data point was not good enuf, now theyre saying 30,000 samples isnt good enuf, so how many billion do you want