Originally Posted by SpankyMcFlych
And so it begins. The neverending posts of anecdotal "I bought 1 drive once and it never had a problem". Honestly who cares if your minuscule sample of seagate drives are just fine. Statistically they're not.
Win or lose (and lets be real, they're probably going to win), seagate loses either way due to the bad publicity in this.
Backblaze's anecdotal evidence of "We chucked a bunch of cheap drives into a extreme environment" has bugger all to do with typical use cases. It's like if you tested the reliability of external drives by shaking them while writing...Sure, you'll more than likely see differences and that some drives are more reliable in that specific environment but I doubt many of us tend to stick 45 drives in a single case which has all sorts of issues with vibrations that even rubber grommets where you mount the drives won't really fix entirely.
...And if you believe their data entirely, even they say the ST4000DM000 is reliable with a 3.06% failure rate over nearly 30,000 drives
. People forget that only HGST beat the 4TB Seagates for reliability which were thrashing the WD 4TB Blacks at one point according to their data iirc.
Originally Posted by Liranan
If all their drives undergo the same stress then it is reliable as they are all in the same environment and there is no bias. Hitachi having the lowest failure rate is due to the quality of their products.
I have a 1TB Hitachi that has been working flawlessly for seven years. If only the same could be said about the WD and Seagate drives I've had in the same period.
That's like if I tested coolers at my workplace (Baker, so flour would go everywhere
) and used it to say what reliability would be in your home...Sometimes drives may not stand up to blatant torture but last perfectly fine in a normal user environment, likewise with other components. Want proof of that? The GTX 570 and 590 would sometimes fail when OCed and stressed with Furmark due to VRMs, while I found the cut down VRMs to be a bad thing personally it means nothing for actual game loads that don't stress the card anywhere nearly as much but can theoretically mean the component won't last as long...Theory is a theory however, not a scientific theory which is practically fact. The issue is that people read about the 1.5TB drives that had firmware issues or 3TB drives with higher than normal failure rates (Which as far as I know has more to do with batches than anything else) and immediately assumed every Seagate drive is bad..It's like putting a WD Green into a highly stressed NAS then saying WD Blacks are crap when it fails.
Originally Posted by Blameless
A less durable part is going to fail more often, even in light loads. Overall failure rates will go down, but a discrepancy will remain.
Stress tests do reveal weaknesses that are generally representative, it's just compresses them into smaller timescales.
It does however depend on a lot of variables that aren't being taken into account. It could be that they're less tolerant of one effect of that kind of stress (eg. Vibration) but it doesn't come enough into play for typical consumer loads that by the time it would actually have a genuine effect, something else has failed on most drives regardless of manufacturer. (eg. Motor bearings to pull an example out of air) If WD had a sticky head issue for example (They have in the past) it also wouldn't show as BBs drives are on 24/7 for most of the year but would certainly show much more frequently in a consumer workload where the drive is span up/down more often.
Another example would be the TLC SSDs, they'll always end up dying quicker in SSD durability tests where they write a tonne of data (Likewise with Sandforce drives that don't get compression although most sites include one with compression and one without) however that means very little if the controller of an MLC drive is much more susceptible to power fluctuations and likely to die earlier if you're in an area where brown/blackouts are common without a UPS. It may be anecdotal evidence but it shows an example for what I'm saying: That was the case for my two old Crucial M4s versus my 840 Evo, they both died from blackouts while the 840 has lasted through a couple yet those tests would show the 840 as dying much earlier because it doesn't have the flash durability of the MLC M4...However even it's flash durability is enough to last me until I upgrade.
Yet another example would be going by an Australian review site for cooler temperatures because we have higher temperatures in general than quite a large number of countries, it may show that some coolers are better in that environment however if you're in a cooler area it may end up being completely different because cooler x isn't as efficient at lower temperatures for whatever reason.
Originally Posted by ZealotKi11er
When I first bought Seagate HDD they had 5 years warranty. They then changed them to 3 years and now only 2 years. Very pathetic. Next time I am getting HDD with 5 years warranty only.
All HDD makers used to be 3 years across the board before 2002, then 3 of the major HDD makers (Maxtor, Seagate and WD) all switched it to 1 year with Hitachi/IBM also doing so a couple of weeks afterwards and only Samsung keeping a 3 year warranty. Seagate made some of their best drives shortly afterwards (The 7200.8s I believe they were called, I still have 3 working ones that all only have a few reallocated sectors but outlived their useful life as 80GB IDE drives) and were the first to go back to a 3 year warranty only to cut it again more recently.
Generally, while warranties can show they don't have faith in a product it really just is a way of cost cutting as the amount of deaths increases exponentially as the drives get older regardless of their reliability. It's consumer unfriendly for the sake of profitability however it's not a clear cut sign of poor reliability...Case in point, my 3TB is showing some signs of wear according to the SMART stats after the Aussie summer but even though my WD Black has a 3 year longer warranty, it fully died during the same summer in the same case with a similar workload despite also being newer. I just grabbed a 4TB WD Elements external and made sure all of my data is backed up because every drive fails eventually and it may not be when you think it'd be.